Skip to main content

Full text of "Work and effort"

See other formats





the Internel 

: Archive 

in 2013  









Copyright, 1947, by 
The Ronald Press Company 

All Rights Reserved 

The text of this publication or any part 
thereof may not be reproduced in any 
manner whatsoever without permission in 
writing from the publisher. 






This book provides a systematic survey of psychological investiga- 
tions concerned with the productivity of men and women at work. It 
attempts to present an evaluated account of the knowledge that has 
been accumulating for many years, to give a survey of current progress 
in this field of investigation, and to put the reader in position to 
realize the need for future development in researches of this character. 

Among the problems treated are training and learning, the control 
of accidents, establishment of pay levels for various jobs, and the 
design of efficient methods of work. Motivation of the worker is 
considered in its relation to problems of efficiency. Because all these 
topics have a common core, they are treated broadly and with regard 
to their general rather than specific implications. 

The book presents conclusions drawn from research in various 
fields. These include a study of the factors of light, heat, and ventila- 
tion, the control of monotony and fatigue, and the provision of con- 
ditions that make for maximum output without excessive effort or 
fatigue. All this information is based upon reliable experimental and 
statistical techniques, to the end that facts, and interpretations which 
are close to the facts, may be provided for the use of industrial man- 
agers and supervisors, personnel directors, and psychologists — in 
short, for all those who are concerned with the setting of standards, 
with installing merit rating and job evaluation systems, and with 
improving the morale and efficiency of workers by the use of tested 
scientific methods. 

In addition to presenting material basically related to the topics 
mentioned above, the book also aims to acquaint the reader with the 
methods of research that led to the findings set forth. This has been 
done so that the reader may determine for himself just how the 
psychologist arrives at any given conclusion. Such a manner of 
presentation should also appeal to students of applied psychology who 
plan to go into engineering, managerial, personnel, or other related 
lines of work. 

In acknowledging obligations to those who aided the author in 
writing this book, the first is inadequately recognized by dedicating 
this work to Professor Madison Bentley. The author is indebted to 



him for not only tuition but also sage advice and helpful guidance 
during the author's initiation into the field of psychology. 

Another important contribution to the work wsls made by the Col- 
lege of Engineering at Cornell University. Some years ago the 
author was asked to join the faculty of that College in order to dis- 
cover, if possible, what contributions a psychologist might be able to 
make to the solution of problems that arise in engineering work. Dean 
S. C. Hollister, responsible for giving the author time and opportunity 
to become acquainted with the engineering problems involved, dis- 
played an active interest in furthering studies in this field. Other 
members of the College faculty contributed in many ways, especially 
Professors John R. Bangs, Jr., C. L. Cottrell, D. Dropkin, H. J. 
Loberg, C. I. Millard, and E. M. Strong. 

Colleagues in the Department of Psychology also gave much help. 
Professor M. E. Bitterman read and helpfully criticized the whole 
manuscript — some parts of it several times at various stages of writ- 
ing. Portions of the material were also read by Professors H. P. 
Weld, H. S. Liddell, and F. L. Marcuse. A former colleague. Pro- 
fessor John G. Jenkins, now of the University of Maryland, first 
introduced the writer to the field covered by this book. Those who 
know Professor Jenkins' work will doubtless find many influences, 
particularly on the general point of view expressed. 

The author is grateful to the many writers and publishers who 
kindly allowed quotations, charts, and tables to be taken from their 
books and articles. These permissions are acknowledged wherever 
the material appears in the text. 

Last of the author's obligations — but far from the least of them — 
he owes to his wife, Dr. Mary Shaw Ryan, who supplied criticism of 
psychological content and literary expression, as well as constant en- 
couragement throughout all the time this book was being written. 

T. A. R. 

Ithaca, New York 
September 2, 1947 


chapter page 

1 Introduction 3 

2 The Meaning of Efficiency or Economy of Work . 20 

3 Measuring the Cost of Work : 1. Metabolism and 

Muscular Fatigue 33 

4 Measuring the Cost of Work: II. Fatigue in Seden- 

tary Work 63 

5 Measuring the Cost of Work : III. Physiological Tests 

OF Effort and Fatigue in Sedentary Work ; Statis- 
tical Indices of Efficiency 95 

6 Specific Factors Governing Efficiency and Produc- 

tivity 124 

7 Work Methods and Efficiency 157 

8 Incentives and Motives 169 

9 ''Nervous" Fatigue and Boredom 188 

10 Time Standards and Rate Setting 209 

11 Merit Rating and Job Evaluation 237 

12 Psychological Problems in Accident Control . . 253 

13 Skill and Practice 297 

Index of Names 313 

Index of Subjects 317 




1 Scatter Diagrams Representing Varying Degrees of Correlation 

Between Grades in Physics and in Calculus .... 14 

2 Relation Between Output and Feelings in Four Forms of Mental 

Work 46 

3 Finger Ergograph as Designed by Mosso 49 

4 Records of Two Ergograms 50 

5 Skin Resistance and Muscle Potentials Following Ergographic 

Work 59 

6 Hypothetical Effects of Fatigue and Practice 67 

7 Composite Hourly Performance Curves for Two Metalworking 

Plants 71 

8 Composite Hourly Performance for Different Classes of Work 

(Eight-hour Metalworking Plant) 72 

9 Hourly Performance Curves for Visual Inspection of Cartridge 

Shells 74 

10 Test Performance Following Three Nights of Diminished Sleep 

(Dotting Test) 82 

11 Test Performance Following Three Nights of Diminished Sleep 

(Associated Words) 83 

12 Action Potentials During Rest and Work (Reading), With and 

Without Distraction 105 

13 The Relation Between Critical Flicker-Fusion Frequency and 

Brightness 115 

14 Key-Pressure in Work During Quiet and Noise .... 126 

15 Average Magnitude of Action Potentials on Successive Days of 

Stimulation 127 

16 American Society of Heating and Ventilating Engineers Comfort 

Chart for Still Air 136 

17 Practice Curves in Computation for Different Instructions . . 172 

18 Performance in Computation Under Two Instructions . . . 173 

19 Average Hourly Output in Western Electric Company Test Room 

Studies 180 




20 Effects of Incentives Upon Different Kinds of Work ... 186 

21 Hypothetical Work Curves 198 

22 Production Curves for a Repetitious Task— Individual Curves 

Compared with Worker's Reports on Boredom .... 200 

23 Production Curves for One Week of Repetitious Work — One 

Worker Who Considers the Work Very Boring .... 201 

24 Fatigue and Personal Allowance Table 226 

25 Standard Time Values for 'Tlace" Corrected for Transport Dis- 

tances 230-231 

26 Example of a Typical Merit Rating Form 239 

27a Suggested Form of Employee Appraisal 247 

27b Inventory of Personnel 248 

28 Hourly Incidence of Accidents in Various Industries . . . 258 

29 Accidents and Production in Eight-hour and Ten-hour Metalwork- 

ing Plants 261 

30 Accident Frequency in Relation to Temperature (Shell Factory) . 263 

31 Accident Disability in Relation to Temperature (Coal Mining) . 264 

32 Driver Testing Apparatus with Moving Road Scene . . . 271 

33 Accident Rates of Dockyard Apprentices, Related to Test Scores 

("Aestheto-Kinetic Tests") 274 

34 Relation of Accidents to Scores on Driver Tests .... 277 

35 Accident Reduction Following the Use of Special Tests (Accident 

Record of 144 Accident Repeaters) 284 



1 Range of Correlations Obtained by Chance for Varying Size of 

Group, Assuming That No Correlation Exists in the Population 

from Which the Group Is Drawn 16 

2 Types of Muscular Work 35 

3 Rough Computation of Mechanical Efficiency 37 

4 Cost of Work in Lifting Loads by Crank 38 

5 Relative Energy Cost of Pulling vs. Pushing ...... 39 

6 Energy Expended in Human Locomotion 40 

7 Kinds of Sedentary (Mental) Work 64 

8 Change in Performance from Morning to Evening ... 84 

9 Proportion of Days Showing Decreased Performance ... 85 

10 Zones of Maximum Comfort 137 

11 Effects of Increased Illumination Upon Industrial Output . . 139 

12 Critical Levels of Illumination Beyond Which There Is Little 

Change in Output 139 

13 Influence of Level of Illumination 140 

14 Load on Fingers During Typing 166 

15 Rank Assigned Various Factors in Morale by Employers and 

Employees 177 

16 Reasons Assigned by 157 Young Men, High School Graduates, 

for Preferring One Job Rather Than Another .... 177 

17 Average Boredom Rating of 68 Workers According to Types of 

Work Curves 199 

18 Average Scores in a Test of Intelligence of Individuals Who Re- 

ceived the Highest and Lowest Assessments for Boredom . . 204 

19 Incidence of Boredom Symptoms in the Case of the Same Group of 

Workers Employed for Monthly Periods in Different Processes 207 

20 Leveling Factors 215 

21 To Illustrate the Application of Leveling Factors .... 216 

22 Time Study of Operators' Performance in a Folding Job . . 220 

23 Mean Cycle-Time, Standard Deviation, and Range . . . 234 




24 Hypothetical Ratings of Three Workers with Equivalent Total 

Scores 242 

25 Factors in Job Evaluation 251 

26 Total Accidents : December 7, 1941 to August 14, 1945 ... 253 

27 Comparison of Morning and Afternoon Accidents . . . 259 

28 Effect of Length of Working Day Upon Increase of Accidents 

in the Afternoon 259 

29 Variation in the Average Number of Hospital Visits per Year 

Among Employees on Eleven Different Jobs 265 

30 Distribution of Automobile Accidents in a Random Sample . . 268 

31 Hypothetical Analysis of Accident Distribution with Two Liability 

Groups 269 

32 Fletcher's Comparison of Test Performance of "Good" and "Poor" 

Drivers 273 

33 Mean Percentage Accident Rate Over the Whole Period of Ex- 

posure of Groups Selected by Different Methods .... 275 

34 Effect on the Subsequent Accident Rate of Removing Drivers 

with Many Accidents in Their First Year or Removing Drivers 

with Poor Aesthetokinetic Co-ordination 275 

35 Vision and Intersection-Type Accidents 276 

36 Correlations with Accident Rate 279 

37 Re-educational Effect of Driver Test Clinics 282 

38 Product-Moment Correlation Coefficients Between Different Types 

of Accident 287 

39 Personal Accidents of Drivers for a Public Utility Company Com- 

pared with Motor Vehicle Accidents of the Same Drivers . . 289 

40 Kinds of Learning 302 



Chapter 1 


Supervisors and managers of industry, teachers, students, and 
workers, all have an interest in the productivity and efficiency of man 
at work. They are interested in providing conditions which are con- 
ducive to maximum output without excessive effort or fatigue. They 
wish to know how important such factors as light, heat, ventilation, 
hours of work, may be in determining performance. They would 
like to know how to control monotony and fatigue, and how to de- 
sign methods of work which are easier and more rapid than those 
which have been used in the past. They inquire most frequently of 
all about such topics as morale and motivation. 

This book is a survey of psychological investigations of these 
problems. It aims to give the student of applied psychology an un- 
derstanding of the basic knowledge in this field which has been ac- 
cumulating during the past few decades. In the relatively short time 
since research upon these problems began, much has been discovered, 
but naturally much remains to be done. We hope that the student 
will not only gain a knowledge of our current progress, but that he 
will also come to a realization of the future developments which are 
necessary in this field before some of the important problems of 
practice can be dealt with on a sound basis. 

We consider it to be the task of the applied psychologist to pro- 
vide information which is based upon reliable experimental and statis- 
tical techniques — in short, to provide facts and interpretations which 
are close to the facts. When the psychologist steps beyond these 
limits, he should make it very clear that he has done so — that he is 
now guessing, using ''common sense" or hunches. 

Complete Solutions vs. Solutions Limited by the Available 
Facts. — The problems with which this book will deal affect every 
human being. Because of their importance it is natural for the 
writer on these subjects to be discontented if he is unable to give a 
well-rounded, easy, and ''helpful" answer to many of the questions 
which arise in the factory, office, school, or home. He is subjected to 




[Ch. I 

great pressure from personnel men, supervisors, and teachers to "give 
an answer we can use right away." The temptation to go far beyond 
the established facts in meeting such a request is very great, especially 
as the layman is not likely to look closely at the evidence upon which 
the recommendations are based, but one we should and wdll try to 

The progress of applied psychology would be much more rapid if 
we disappointed industrial managers and supervisors more fre- 
quently, pointing out that we would have to guess, with little guaran- 
tee that our guesses would be better than anyone else's. Industry has 
had to put engineers to work solving its physical problems and filling 
the gaps in available information. There is no reason why industrial 
and business men should expect the universities to fill in these gaps 
in psychological knowledge without support, and they would not do 
so if they realized what is involved in the questions they raise. 

Let us take a simple problem from our own field to illustrate the 
point. Suppose we are asked to recommend the best temperature to 
be maintained in an office building, from the point of view of efficient 
mental work. Someone has mysteriously arrived at the conclusion 
that 68 °, 50 per cent relative humidity, is the best and most healthful 
condition for sedentary people. That recommendation is given again 
and again without any reference to the fact that there is no discover- 
able and adequate experimental basis for the recommendation. 
Would it not be better to point this out, and refuse to give a recom- 
mendation until those who ask the question are willing to .support an 
experimental program for seeking an answer to it ? 

The point of view expressed here is not always accepted, even by 
psychologists. There are those who feel that the psychologist should 
try to give some definite recommendation even in those problems in 
which it is necessary to *'go out on a limb" and make statements 
based upon personal opinion. The justification which they use is that 
the business man, the teacher, or the industrialist must make a de- 
cision about the question. If the psychologist will not give him a 
definite answer, he will use his own judgment, or he may become the 
prey of quacks, or other unscrupulous persons who claim to solve all 
problems by some method completely unfounded in scientific investi- 
gation. Some psychologists, concerned about this state of affairs, 
say, in effect, "Would it not be better for the psychologist to try to 
give an answer to the best of his ability ? The psychologist may make 
mistakes, but his conclusions will be based upon the best available 
evidence, and certainly would not be based upon theories which are 
entirely false and misleading." 

Ch. i] 



This argument may have force when it refers to the functions of 
the psychologist in private consultation on some specific problem, 
although even there it is questionable whether the opinions of the 
psychologist are to be given more weight than those of an experienced 
supervisor or executive. In such consultations, the main advantage 
which the psychologist has is that he is outside the situation and can 
look at the problem more impersonally. In many problems, the ac- 
tual facts which he can use in drawing his conclusions will be less 
important than this impersonal, objective approach. 

A book aiming to present some general field of applied psychology, 
and attempting to follow the procedure we have just outlined (the 
procedure of being ''practical" at all costs), would have several serious 
defects. If the book is limited to recommendations for practice, the 
reader has no way of judging which recommendations are based upon 
a wealth of research information and which are based almost entirely 
upon guessing. Just as important as this defect is the fact that the 
book must be either an encyclopedia of hundreds of specific problems 
and their solution, or it must attempt to present so-called general 
"principles" which are presumed to be general enough to apply in a 
variety of situations. In the former case, the possibilities of specific 
situations and problems are never exhausted, and in the latter case 
the general "principles" themselves are products of judgment, and 
further judgment must be exercised in their application. 

It should be just as easy for the student of psychology or the prac- 
tical man of afYairs to learn something of the facts and methods by 
which the psychologist works, as it is to learn the recommendations 
found in books of either of the types described in the preceding para- 
graph. The results should be more effective in practice. The man of 
affairs would be less tempted to succumb to the salesmanship of the 
quacks if he knew something of the errors in their methods. 

Need for Understanding the Methodology of Psychological Re- 
search. — Even if we limit ourselves to a discussion of facts and 
methods whose value has been the subject of controlled research, 
there is still another choice which must be made. We may present 
only the conclusions to be drawn from the research which has been 
done in each field, or we may try to show the reader how those con- 
clusions were derived by giving him a summary of the research 
methods and results upon which the conclusions are based. The first 
alternative is frequently adopted in texts upon applied psychology or 
personnel management, apparently on the assumption that those who 
are not planning to carry on research of their own do not need the 
background of methods. 



[Ch. I 

There are several reasons why we disagree with this latter point 
of view, and why we believe that some treatment of the methodology 
of research is essential, even if the presentation is not specifically de- 
signed for the student who will engage in research of his own. Even 
the industrial manager or engineer who is immediately concerned 
with practice would profit, we believe, by a knowledge of research 
methods and specific results in applied psychology. Perhaps he would 
profit even more from this kind of information than he would from 
general conclusions and recommendations. Here, in brief, are some 
of the reasons : 

1. Conclusions are often drawn with reference to a certain state of 
affairs, or conditions which exist now. When new situations 
arise, the conclusions may not be applicable at all, while certain 
of the facts would be reinterpreted with reference to the new 

2. The practical man of affairs may feel that the psychologist is 
being too cautious and limiting his conclusions too much — hedg- 
ing them about with qualifications and provisos. If the facts 
themselves are available, he should see that the only way in 
which the conclusions can be broadened with assurance is by 
more extensive research. If the reader wishes to draw wider 
conclusions he may do so — at his own risk and with better appre- 
ciation of the risk. 

3. If the layman can arrive at an appreciation of the care which is 
necessary in research to arrive at reliable and valid results, he 
will become aware of the unreliability of common sense and 
experience as means of solving practical psychological problems. 
For example, if he sees how important such factors as suggestion 
may be in determining the effects of any given factor, he will 
look more critically at his own experience and at the anecdotes 
which he hears from others. 

4. Many students of applied psychology, while they plan to go into 
engineering, managerial, personnel, or other similar types of 
work, will frequently become involved in some way in psycho- 
logical research. They may have to deal with some problems 
themselves, they may have to decide when the services of a 
specialist in a particular kind of research are needed, or they may 
have to evaluate the results of research carried out by others. 

In this book, therefore, we shall take the position that one cannot 
get the most out of the results of applied psychological researches 
without knowing something of the method by which the facts were 

Ch. i] 



derived. It is not enough that the writer aims to draw his conclusions 
from facts, rather than from opinion and speculation. The scope and 
meaning of the generalities are not clear without the research back- 
ground. The engineer who has to work in a technological field is ex- 
pected to know something of the fundamental research methods 
underlying the procedures which he applies. Similarly, one who 
wishes to deal with practical problems of human behavior in a techno- 
logical fashion should have a background in psychological research. 

Major Problems of Work 

Before going into detailed analysis of any one phase of work and 
efficiency, it would be well to investigate the scope of the field which 
we propose to study. Let us consider the broad problems involved, 
and their general meaning. It will be seen that these problems, which 
appear at first quite distinct, have a common core which will be made 
the central theme of this book. 

1. Problems of Efficiency. — Here are included such questions as 
these : ''What is the best way to learn to operate a typewriter ?" *Ts 
it better to use a great deal of specialization, or let each worker per- 
form a number of tasks ''How frequently should the workers on 
a certain job rest, and what should they do during the period of rest?'* 
"Would it be worth while installing better lights, an air-conditioning 
system?" "Would I be just as well off with six hours of sleep every 
night instead of eight or nine — am I wasting time in sleeping?" 

This list could be extended indefinitely, of course. It is also evi- 
dent that many of these questions are complex, and so cannot be 
answered fully as yet. Before we can even seek an answer, however, 
it will be necessary to clarify the meaning of many such questions. 
When we ask, "What is the best way to do a certain thing?" we may 
get in reply a counter-question, "The best way for whom?" Would 
the answer be the same from the point of view of the worker, of the 
employer, or of the community at large? Even more generally, we 
may ask, "The best way from what point of view?" Do we want to 
seek the method which is the best from the point of view of rate of 
production, amount of fatigue per unit of work, or interest in the 
work? In brief, when we consider questions of efficiency, we shall 
have to consider what we mean by efficiency, and also whether dif- 
ferent points of view will lead to the same result. Until we do so, we 
shall not know what we are looking for. 

We shall consider efficiency in terms of a relationship between the 
cost of work to the individual (effort, fatigue, energy expended, etc.) 



[Ch. I 

and the outcome. If cost of work is constant while the output in- 
creases, we shall speak of an increase in efficiency. If the output is 
increased but the cost is unknown, we shall not be justified in speaking 
of an increase in efficiency. 

Terms like effort and fatigue are very familiar and are readily 
understood in the above definition. We shall find, however, that the 
measurement of effort and fatigue requires something more than a 
general understanding of the terms. In fact, these are key problems 
of the psychology of work. 

2. Problems of Motivation. — One group of questions under this 
heading concerns individual differences in motivation : "Why does A 
get so bored with his work, while B is much more interested?" 
**What can be done to increase A's interest, or how can another job 
be found which will appeal to him more?" Another set of questions 
relates to group or average motivation : "What inducements can be 
offered to make a suggestion-system more effective?" "What kind 
of campaign will induce workers to be more careful about leaving 
things about the floor?" "How can the productivity of this plant be 
increased without a greater working force — i.e., how can the workers 
be induced to work harder? What kind of bonus system would be 
most effective?" "How can we persuade the stenographers to remain 
on their jobs and stop looking for factory jobs?" 

There is a very simple relationship between problems of efficiency 
and problems of motivation — at least in the conception, if not in the 
solution, of the problems. In problems of motivation, we are basi- 
cally concerned with raising the level of effort that the worker puts 
into his work, and thus increasing his productivity. In the problems 
of efficiency which were discussed above, the problem is to increase 
the results which are obtained from a given amount of effort — to see 
that the effort is expended more economically. 

3. Special Problems. — There are a number of problems which 
are already included under the headings of efficiency and motivation. 
Because of their unique characteristics or their special importance, 
however, they are usually given separate mention : 

{a) Training and learning. 

(&) Accident control. 

(c) Establishing relative pay levels for various jobs. 

{d) Selecting workers. 

Of these special topics, the last, selecting workers, has received so 
much elaboration in recent years, and there are so many specialized 

Ch. i] 



techniques available, that in this book we shall mention it only in 
passing. Our interest will be mainly that of relating the problem to 
other aspects of efficiency and motivation. 

Topic (c) is another instance of the great importance of measure- 
ment of effort. Here we come immediately to the question, "For 
what is the worker being paid?" It is obvious that he is being paid 
for his skill, knowledge, and the effort he expends upon his work. In 
a sense, he is paid for past and present effort, plus the aptitude which 
he brings with him from his family inheritance and early upbringing. 
This is an extremely simplified statement, but when we come to an- 
alyze the problem of establishing pay levels we shall see that the 
measurement of effort is involved in a number of complex ways. In 
fact, there are several methods that are spoken of as ''scientific" in 
this field which do not deserve that label because they are not based 
upon any sound method of evaluating effort. 

Methods of Gathering Facts and 
Determining Their Interrelationships 

As soon as we enter any field of psychology we are confronted 
with a maze of opinions, statements, and claims. What criteria shall 
we use in sorting these statements and in determining which are 
worthy of serious consideration, which are untrue, and which require 
further checking? For example, which of the following statements 
would you accept as factual ? 

1. Women are better adapted to repetitious work than men. 

2. Women cannot work as long hours as men. 

3. Persons of normal or high intelligence are very unlikely to do 
well at simple manual and mechanical tasks. 

4. The primary incentive factor in industry today is the pay rate. 

5. Loss of a night's sleep has a serious effect upon performance of 
a job the next day. 

Some of these statements refer to a state of affairs which is sup- 
posed to exist. In other cases (for example, Statement No. 5) a 
cause-and-effect relationship is involved. In the latter case we must 
determine a relationship between the facts, as well as establish the 
facts themselves. Let us first consider what would be necessary to 
establish our facts, and then come back to the further problems of 
cause-and-effect or other relationships between the facts. 

The statement that women are better adapted than men to repe- 
titious work is based upon the opinions and reports of personnel 



[Ch. I 

administrators and supervisors. Does this justify its acceptance as 
an established fact? It is no reflection upon the veracity of these 
observers if we suggest that they might be wrong in their opinion. It 
it not a question of honesty, but rather a question as to whether it is 
possible to arrive at such a conclusion on the basis of common, every- 
day experience. Such statements have been made with assurance in 
the past and have later been proved to be wrong. We no longer be- 
lieve that the earth is flat, but present-day psychological beliefs are in 
many ways comparable to the knowledge of geography and astronomy 
which was available to Columbus. 

A very recent example illustrates the danger of trusting to this 
method of common observation for our "facts." Management of a 
certain concern stated that in that organization the worst day of the 
week for absenteeism was the day after payday. Study of the sta- 
tistics of absences of that organization, however, showed that Mon- 
day had a considerably higher rate of absences than the day that had 
been called the worst of the week. If a mistake of this kind, dealing 
with a simple question of frequency of absence can be made, how 
many mistakes must appear when we deal with something less tan- 
gible; such, for example, as the degree of adjustment to repetitious 
work I 

This example of absentee rate brings out clearly what is needed to 
establish the correctness of other statements — measurement or count- 
ing instead of unreliable estimation. We must have some measure of 
degree of adjustment to routine work, or we must at least be able to 
classify men and women into groups (high, medium, or low). This 
measure of adjustment must be applied to a representative sample of 
both men and women, working on a variety of jobs. In order to 
make the general statement, we should have to be certain that we have 
studied representative routine jobs and, of course, the same jobs for 
both men and women as well as a representative group of men and 
women. Finally, we need to know the statistical significance of the 
difference, as well as the size of the difference and its practical im- 
portance. We need, in short, a statistical analysis of the results. 

Statistical and Practical Significance. — To a scientific worker, 
statistical analysis means much more than mere computation of av- 
erages. Much more important is the interpretation of the results 
from the point of view of chance. When we say that a result is 
statistically significant, we mean that we have good reason to accept 
the result as more than mere luck, that we could repeat the experiment 
and get similar results again, that the result is dependable. 

Ch. i] 



Suppose, for example, that we had studied 100 men and 100 
women working upon a dozen different routine kinds of work. We 
might get any one of the following results : 

(a) A large but wsignificant difference in adaptation to work. 
That is, while the difference between the averages appears sub- 
stantial, there is so much individual variation among both men 
and women that we should have to have a much larger group 
in order to determine whether any real difference exists. 

(b) A large, significant difference. That is a difference which 
could be depended upon, which would again appear if the study 
were to be repeated. 

(c) A small, significant difference. This would also indicate that 
there is some real difference between men and women in this 
respect, but that there would be a considerable amount of over- 
lapping. A substantial number of men would be better than 
many of the women, and, conversely, many women would fall 
below the men's average. 

A significant difference is not necessarily of great practical useful- 
ness. In case (c) above, we might have only a slightly better chance 
of getting a satisfactory worker if we hired a woman instead of a 
man. Statistical significance depends not only upon the size of dif- 
ference but also upon the number of cases studied and the amount of 
individual variation within the groups. A very slight difference be- 
tween the averages of two groups might be significant if there were 
a million people in each group, although individual members of each 
group are spread over a wide range. 

We must establish the statistical significance of any result before 
we are justified in drawing any conclusions at all. Once it is estab- 
lished, we must still determine the practical meaning of the result. 

Two common errors which are found in reports of investigations 
of practical psychological problems involve the following points : 
First, there are many investigations carried out by practical men of 
affairs who believe that statistical analysis ends with the computation 
of averages. Obviously, in the light of what has just been said, no 
conclusions at all are possible. The second error is sometimes over- 
looked even by writers who have some knowledge of statistical 
method. This consists in determining the statistical significance, but 
failing to evaluate the findings from the point of view of practical 
applicability. The statement quoted in the list above, to the effect 
that intelligence is related to ability to adjust to repetitious work, is 
based upon such a failure. Barely significant and quite small differ- 



[Ch. I 

ences were obtained in the investigation. Yet some books on per- 
sonnel management regard these results as showing the practical value 
of selecting routine workers on the basis of intelligence level. Actu- 
ally, the overlapping between the groups was so great, and the number 
of different jobs studied was so small, that the significance of the 
difference is not enough to establish intelligence tests as a useful 
selective tool. 

A simple numerical illustration should make the latter point clear. 
In a hypothetical study, the proportions of men and women who are 
seriously affected by boredom in their work are as follows : 

If the two groups are sufficiently large, this difference could be 
highly significant in a statistical sense. This means that further 
studies with large numbers of individuals would also obtain differ- 
ences in the same direction. The decision to eliminate men from this 
kind of work, however, would result in only a small decrease in the 
percentage of workers who are bored with the work. Many other 
considerations would have to be taken into account before we could 
decide whether the elimination of male workers would have any real 
advantage at all. 

Relationships Between Variables. — Two general kinds of inter- 
relationship between events are of importance in applied psychology. 
We may be interested in a relationship between two variables because 
we wish to predict one of them from the other. Thus, we may wish 
to predict performance in a certain college course from an individual's 
performance in a previous course. In this case we are not concerned 
with the reasons for the relationship. We are satisfied if some pre- 
dictable relationship exists. In other cases, however, we are in- 
terested in the relationship because we want to know what kind of 
relationship exists. We wish to know in what way one variable de- 
pends upon the other. Thus, if performance in a course in physics 
is predictable from performance in an earlier course in calculus, we 
may want to know whether the dependence is due to the need for the 
material learned in the calculus course, or whether the calculus grade 
is merely an indication of a special aptitude of the student which de- 
termines good physics grades regardless of whether he has studied 
calculus or not. In other words, does performance in physics depend 
upon the calculus learned, or are the two variables related through 

Men . . 


Per Cent 
Seriously Bored 



Ch. i] 



some third factor ? In both cases we could predict the second variable 
from the first, but our interpretation of the facts and other applica- 
tions would differ in the two cases. 

We have a variety of measures of statistical relationships between 
variables at our disposal. None of these, however, can do more than 
measure the predictability of the relationship; they do not indicate 
the nature of the dependence, the presence or absence of third factors, 
or anything of that sort. They merely tell us how accurately one vari- 
able can be predicted from the other. 

Determining the predictive relationship is, nevertheless, a step 
which is useful in itself, and which must be part of any analysis of the 
dependence of variables. It will be well, therefore, to become fa- 
miliar with, or to review, as the case may be, the important tools that 
are used for this purpose. 

The most commonly used measure of predictability is the coefficient 
of correlation. Its use is so widespread that no one can hope to read 
very far in serious psychological literature without at least a general 
understanding of its meaning. The coefficient of correlation (com- 
monly designated by the letter r) is an index which varies from 
— 1.00 to + 1.00. To return to the illustration of the relationship 
between calculus and physics, we should investigate that relationship 
by obtaining the records in both courses of a large number of stu- 
dents. We might plot the points representing the individual students 
upon a graph like those shown in Figure 1, where the x axis repre- 
sents performance in calculus, and the y axis represents performance 
in physics. If all the points fell on a straight line like that in Figure 
la, we would have a perfect correlation (1.00), and we would be able 
to predict physics grades exactly, once the calculus grade were known. 
If the points scattered, as in Figure lb, we could not predict from one 
variable to the other any better than we could if we drew numbers 
out of a hat. This is the condition expressed as a coefficient of 0.00. 
A minus sign for a coefficient merely indicates an inverse relation- 
ship. In this case a negative correlation would mean that physics 
grades are lower, the higher the student's grade in calculus. Figure 
Ic represents the "scatter diagram" which would be obtained in a 
case of — 1.00 correlation. 

Correlations of 1.00 are never obtained in practice because of vari- 
ations due to chance (any unpredictable variable) and those due to 
other factors which are not under control or study. For example, 
the number of hours of work the student is taking at the time of the 
course in physics may not be the same as that taken at the time of the 
course in calculus. We must, therefore, be able to deal with cases in 



[Ch. I 

which there is a general, but not exact, relationship between the vari- 
ables. This situation is expressed by a correlation coefficient between 
and 1 (with either a positive or negative sign, depending upon the 
direction of the relationship). Figure Id represents results which 
would be expressed by such a coefficient. 





dT'' high, positive 

Figure 1. Scatter Diagrams Representing Varying Degrees of Correlation Between 
Grades in Physics and in Calculus 

Each point represents the pair of grades obtained by one individual. In (a) and (c) all 
points fall on the dotted line. 

It should be noted that the symbol r is restricted to the measure of 
linear relationships between two variables. We can write a simple 
equation expressing the average relationship between the two vari- 
ables. Thus, we might find the following relationship for our ex- 
ample : 

Grade in physics = .5 X Grade in calculus + 50 

Ch. i] 



The value of r will represent the accuracy of predictions made on the 
basis of this equation. Knowing the value of r and applying a few 
simple formulas, we can determine the amount of error which can be 
expected in using calculus grades as predictors of success in physics. 
The more closely r approaches 1.00, the smaller the errors which will 
be made in predicting. 

A question which is frequently raised is, How large a value of r 
is needed for ''good" prediction — for predictions which will be prac- 
tically useful. This is not, however, a question which can be an- 
swered. The practical value of a given degree of relationship 
expressed by the index r varies with the circumstances. It depends 
upon whether we are concerned with making accurate predictions for 
a given individual, or with improving the average accuracy of predic- 
tion for a large group of individuals. Obviously, the closeness of 
relationship must be the larger in the former case than in the latter. 
The practical value of a relationship depends also upon the kind of 
prediction which is to be made. If we are to predict only whether a 
person will pass or fail, we do not need so large a correlation as we do 
for predicting whether the individual will get an A, B, or C, etc. For 
these reasons, and for several others, there is no absolute interpreta- 
tion of a given value for r in terms of practical value in prediction. 
Usually, however, the writer of psychological articles shows the kind 
of prediction in which he is interested, and makes an interpretation 
of r upon this basis. 

The problem of significance arises again when we deal with the 
coefficient of correlation. In studying the relation between two vari- 
ables, we must always deal with a sample of observations. A certain 
degree of correlation might appear in a given sample, even though 
there is no relationship between the variables in the long run — on the 
average for all possible observations. Thus, if we were to take the 
records of only ten students for our example of calculus and physics, 
we might, by chance, find a close correlation between the two grades 
in this group, of ten students. Ten other students, on the other hand, 
might show no relationship at all. Table 1 shows the amounts of 
correlation which may be obtained by chance with varying sizes of 
sample. To be significant of any general relationship, a correlation 
must fall outside these limits. 

If, then, we were to observe a correlation of .50 in a group of 
twenty-five cases, we could draw no conclusions. There might be a 
relationship which would appear again in other groups, or there 
might be no such relationship at all. All we can do is to make further 
Studies to find out. If the correlation of .50 appeared in a group of 


a hundred cases, however, it would indicate that some real and de- 
pendable relationship exists. It might not be exactly .50, but we can 
be quite sure that some positive relationship exists, and so we speak 
of this result as a significant correlation. 


Range of Correlations Which May be Obtained by Chance for 
Varying Size of Group, Assuming That no Correlation Exists 
IN the Population from Which the Group is Drawn 

Number of Cases 

Chance Range 


± 1.00 


± .94 


± .60 


± .42 


± .30 


± .09 


± .03 

We may also use other measures of correlation, measures which 
apply when there is a curved relationship between the variables, in 
place of the linear relationship we have discussed in showing the 
meaning of r, and also measures which allow us to relate several 
variables together. For example, we can measure the predictability 
of a certain variable from two or more other factors. To follow^ our 
previous example, we might be interested in predicting physics grades 
from both calculus grades and grades in elementary chemistry. The 
multiple correlation (i^) is similar to the simple linear correlation in 
every phase of its interpretation, except that it shows the accuracy 
of predicting from several variables at once. 

The Dependence of Variables. — The statistical measures de- 
scribed in the preceding section tell us whether or not some relation- 
ship between variables exists, and how accurately we can predict one 
variable from others. Now we wish to determine what kind of re- 
lationship exists, not what kind of mathematical relationship we have, 
but what kind of psychological or physiological dependence exists. 

To accomplish this aim, we must pass beyond the simple intercor- 
relation of variables to some form of experimental control of other 
factors. To return to the example which we used throughout the sec- 
tion on correlation, suppose we wish to demonstrate that the physics 
grade is dependent upon the material which is learned in calculus, 
independently of the special aptitude which might show its effects in 
both activities. In order to do this, it is necessary to analyze the rela- 

Ch. i] 



tionship between physics and calculus within groups of equal apti- 
tude. It becomes necessary to measure this third factor of aptitude in 
order to control its effects upon our results. Until we are able to find 
out what this aptitude is, and to develop some means of measuring 
it, we shall be unable to solve our problem. 

If we are not certain what the additional variables may be, as in 
this case we may not be certain what kind of aptitude could be the 
link between the two performances, we can resort to a mass control 
of a number of additional variables which are likely to include those 
which are critical to the problem at hand. In our example we may 
control intelligence, numerical ability, mechanical comprehension, and 
a number of other factors which are already measurable by tests, even 
though we do not know whether, or how, they are involved in either 
calculus or physics grades. The procedure would be as follows : 

1. Administer all the aptitude tests to a group who had high achieve- 
ment in calculus, to a group who had low achievement in calculus, 
and perhaps also to a group who had never taken calculus (as- 
suming that the latter are admitted to the course in physics). 

2. Equalize all three groups with respect to their aptitude test per- 
formance. A number of devices may be used to ensure that the 
groups are equivalent in this respect; one common method is to 
match each individual in one group with an individual in another 
group who has nearly the same aptitude test scores. 

3. Compare the performance in physics with the performance in 
calculus for these equalized groups. 

If a previous correlation in uncontrolled groups no longer appears 
in these matched groups, we could be sure that the predictive relation- 
ship between the two courses depends upon additional variables con- 
tained in the aptitude tests. If some correlation still remains, two 
interpretations are still possible. Either the knowledge of the calcu- 
lus contributes directly to the physics grade, or we have failed to 
control some additional variable — our tests have not covered all 
phases of aptitude, or factors like interest and motivation have been 
left out of account. 

Before we can draw conclusions about the dependence of variables 
with certainty, therefore, we must control or hold constant all po- 
tential third factors. 

Let us consider another example to illustrate another way in which 
the same general goal may be accomplished. On page 9 we con- 
sidered some statements whose accuracy we need to check before they 
are accepted. One statement read, "Loss of a night's sleep has a 



[Ch. I 

serious effect upon performance of a job the next day." It is a state- 
ment which would probably have many endorsers among laymen. 
Notice that the statement is more than a statement of predictive re- 
lationships. It does not say merely that a person who had little sleep 
is unlikely to perform well the next day, it says that the loss of sleep 
is responsible for the poor performance. It is, in other words, a 
statement about the dependence of these two variables. 

We might question first whether the predictive relationship is itself 
true, but let us suppose that for purposes of illustration we grant that 
it is correct. Suppose, in other words, that, on the average, persons 
with little sleep on a given night do perform poorly the next day on 
their jobs. When that has been determined, we have still no conclu- 
sive evidence that the loss of sleep is responsible. People who do not 
get a full night's sleep frequently have attended parties, consumed al- 
cohol, danced, smoked more than usual, consumed unusual food, and 
so on. We must somehow eliminate these factors before we can de- 
termine the effect of the sleep loss proper. 

Someone might suggest that these factors could be eliminated by 
studying only those who lost sleep because they had insomnia and 
"just couldn't sleep well." This procedure would evidently eliminate 
the factors we have mentioned so far. There are still others to be 
considered, however. Perhaps the person who didn't sleep well is 
suffering from a temporary upset in health which would also affect 
his work. Many persons worry about losing sleep, and expect the 
loss to have an adverse effect upon their work. 

In many fields of psychological investigation, variables like the 
last are the most difficult to control and in some cases they have never 
been satisfactorily eliminated from the situation. In this respect, a 
laboratory study is likely to be more satisfactory than an investigation 
carried on directly in the shop or factory. The laboratory subjects 
frequently are less likely to be biased in their expectations, their other 
activities can be kept under observation, and their performance more 
carefully measured. This is not to say that an experiment in a lab- 
oratory is always to be preferred to an investigation in a more 
natural setting. In some instances the laboratory situation itself pro- 
duces distorting variables which destroy the accuracy of the experi- 
ment, or make it difficult to apply the results to practical situations. 
Nevertheless there are many problems which require laboratory study 
in order to eliminate variables which are uncontrollable in the natural 

In this introductory discussion we shall not attempt to catalogue 
all the devices which may be used to control the effects of extraneous 

Ch. i] 



variables. These examples should suffice to indicate the importance 
and the difficulty of the problem. As we discuss the various specific 
problems of efficiency, we shall become more intimately acquainted 
with many of these devices. 

This discussion of the steps necessary, first, to establish a predic- 
tive relationship, and, secondly, to determine the interdependence of 
variables, should make it clear why so little faith can be placed in 
statements and claims based upon common sense or common observa- 
tion. These common observations may be wrong even when it is a 
matter of simple statistical or predictive relationships. How can we 
be sure, from simple qualitative observation, that students who do 
well in calculus also do well in physics, or that a person without much 
sleep does not maintain his normal production the next day? We 
should certainly have difficulty in estimating the closeness of the re- 
lationship, and our observations would be so limited that the factor 
of chance could never be accurately estimated. 

If we would frequently be in error about the existence of a statis- 
tical correlation between two variables when we base our opinion 
upon common observations, how much more opportunity for error 
is there when we are concerned with dependence of variables as well 
as predictability? 

Chapter 2 


Since the study of efficiency is to be the central topic of this book, 
and since it is the ultimate concern of almost any investigation in the 
field of applied psychology, it would be well to consider the meaning 
of the term and the implications of this meaning in some detail. In 
later chapters we shall show how research methods are applied to 
scientific and practical problems of efficiency arising in the factory, 
the office, and the school. In these later instances we shall find that 
the measurement of efficiency is only partially realized or approxi- 
mated. The measurement of efficiency of the human organism_at 
work represents a goal, or an ideal, of applied psychology which is 
never quite reached in practice. It is for this reason that it is espe- 
cially important that we have a clear conception of the nature of this 
goal or ideal. 

To avoid confusion, terms which are held in common by the vari- 
ous sciences should have similar if not identical meanings. As used 
in physics and mechanics, the term "efficiency" refers to the propor- 
tion of the energy input of a machine which is recovered in useful 
work. A similar ratio of output to input is essential if the term is to 
be applied to the human organism. For some purposes we can use an 
identical definition in physics, in physiology, and in psychology. For 
other purposes, however, we shall find it necessary to broaden the 
term somewhat beyond the strictly mechanical usage. 

This broader usage of the term would make the phrase ''efficient 
performance" equivalent to ''economical performance." That is, we 
should consider not only immediate energy consumption, as we do 
when measuring mechanical efficiency, but also the over-all cost of 
work. The buyer of a motor car is interested not only in its fuel 
consumption (the mechanical efficiency of the engine) but in the cost 
per mile of operation, including initial expense, depreciation, and re~ 
pairs, along with the fuel. Similarly, for practical psychological prob- 
lems, it may not be enough to know the mechanical efficiency of the 
body in a given task. Other elements of cost may be much more 
important to economical operation. 


Ch. 2] 



Note that even though we are not using the term efficiency in its 
exact physical meaning, we are, nevertheless, retaining the important 
feature of a ratio between output and input. We have simply broad- 
ened the concept of input beyond the energy immediately consumed. 
The other elements of cost also ultimately involve energy, but it is 
not convenient or possible to express them all in energy terms. 

Misuses of the Term "Efficiency." — In common speech, the term 
''efficiency" is used in many vague and often misleading senses. 
Some of these misuses of the term have led to misunderstandings 
and errors. The commonest confusion is between efficiency and rate 
of performance. Even some psychological texts have a chapter en- 
titled ''efficiency" which begins with a careful definition in terms of 
output and input, but confuses efficiency with rate of performance 
before the chapter is over. 

If the rate of performance is increased, efficiency may go up or 
down, or it may be unaffected, depending upon simultaneous changes 
in the cost of work. If we do not know anything about the cost of 
work to the individual, we are safer and more honest if we simply 
report the facts in terms of rate of performance, and avoid using the 
term efficiency. 

Input, Cost, Energy, and Effort. — In addition to the energy trans- 
formation directly measurable during performance (metabolism), 
the input or cost of work includes such factors as fatigue, the time 
of the individual, effects upon his health, his personal adjustment in 
society, his ability to enjoy leisure time and recreation, and possibly 
other factors. In most instances, these factors are more important 
than energy. We need a convenient term to cover this varied list of 
items. The term energy is frequently used in common speech to 
refer to various aspects of the cost of work, but this is likely to lead 
to confusion with the strictly physical terminology involved in physio- 
logical studies of metabolism, as will be seen later. The term input 
is often closely identified with fuel consumption, and cost might imply 
a monetary basis of evaluation. As applied to the human organism, 
these implications should be avoided; they will be avoided in this 
book. Input and cost will therefore be used as equivalent terms, with 
a very general meaning. Included in input will be any adverse effect 
of the work, anything which the individual expends or uses up in 
carrying on the activity under study. 

Input involves many different factors or elements which will be 
considered in greater detail later. In order to avoid confusion, how- 
ever, we may note how two of the important elements of input will be 



[Ch. 2 

defined in this book. Energy expenditure will be used to designate 
only those aspects of activity which can be expressed in terms of 
physical units — calories, watts, etc. It will refer only to that phase 
of input which is taken into account in the physical meaning of the 
term efficiency. Effort will be defined in a way which is somewhat 
different from common usage. This is necessary because the term, 
while used constantly in common speech, has no clear meaning in that 
usage which would permit exact study or measurement. The most 
useful definition of effort would be that it describes the relationship 
between actual rate of performance and the capacity of the individual 
at a given time. Effort involves energy, but it also involves other 
things, so that the two terms should not be equated or confused. For 
example, there are many tasks of the organism which involve very 
little expenditure of energy, but which may be carried on at a very 
high level of effort; (for example, mental arithmetic). This defi- 
nition of effort includes many of the usages of common speech, but 
it makes the meaning more exact and shows us what is required if we 
are to measure effort.^ It will also appear later that this conception 
of effort provides us with a key to the study of the relationship 
between motivation and efficiency. 

The Importance of Measuring Efficiency. — Some may object at 
this point that any supervisor, industrial manager, or even a teacher 
is mainly concerned with output and results rather than with effi- 
ciency. They may also say, as they discover some of the difficulties 
of studying efficiency, that the notion complicates matters too much ; 
that everything would be simpler if we just forgot about it. Prob- 
lems are not solved by ignoring them, however, and some of the 
problems we should be ignoring would be extremely important. 

Let us consider some of the places in which an adequate analysis of 
efficiency is required. In some cases the reader will be surprised to 
find wide ramifications of this problem. 

1. The problem of wage and time standards in industry. The ideal 
here would be to establish payment methods so that each task would be 
paid for in proportion to the effort required to do it. As we shall see 
when we discuss time study and related problems, this is essentially what 
various methods now in use are designed to do. They are little more 
than rules of thumb, however, and will remain so until our studies of 
effort and the cost of work have progressed further. 

^ This meaning of the term has been used explicitly in a recent investigation of 
college aptitude testing, with results which show considerable promise. See P. R. 
Sappenfield, Prediction of college scholarship for groups having effort indices of 
restricted range, Journal of Applied Psychology, 1943, 27, 448-451. 

Ch. 2] 



The relationship can be seen best in a simple example. Two men are 
working at jobs requiring similar training and schooling, and involving 
about equal risk or danger. In other words, their ''investment" is about 
the same. One man produces 100 pieces of the part he is working on. 
The other has a different task at which he is able to produce 50 pieces 
per hour. What should be their relative pay? If we wish to use a 
system of rewards for hard work, we must raise the question of effort. 
Does it take more effort to produce 100 pieces on the first job in com- 
parison to the 50 pieces on the second job? Various crude solutions of 
the problem are now in use, but their validity is questionable because 
they have no satisfactory way of estimating effort. (See Chapter 10.) 

2. Methods of increasing productivity of a given department, plant, 
or industry. If inadequate consideration is given to the input side of 
worker performance, a method of work which at first appears to raise 
production may eventually reduce it. This is due to the possibility that 
some of the cost does not show its effect for some time. A method 
which is to be permanently effective must consider these long-term ele- 
ments of cost and try to find out about them in advance. Otherwise, the 
only way they will be discovered will be through an expensive process 
of long-term trial and error. 

3. Hidden expenses. The overhead of an industrial enterprise is 
definitely affected by turnover, accident rates, discontent, and complaints. 
These, in turn, may be the result of methods which lead to high pro- 
ductivity, but relatively low efficiency of human performance. In other 
words, inefficient performance of the worker is, in the last analysis, a 
condition of inefficient enterprise even though this may not appear at 
once under superficial examination. 

Applying the Concept of Efficiency: An Illustrative Example. — 

In order to make the notion of efificiency more concrete, let us con- 
sider a single problem which occurs in modern industry. The exact 
methods of solving this problem must await later consideration, when 
we have studied in more detail the various experimental methods of 
measuring the cost of work to the individual. At present we shall 
merely examine the implications of the problem from the point of 
view of efificiency as the term is used here. 

As our example, we chose the general problem of specialization 
and its effect upon the efificiency of the worker. Of course, speciali- 
zation is likely to have different effects for different conditions and 
types of work, so that we shall have to make the problem more spe- 
cific by considering its effect upon a single task and comparing two 



[Ch. 2 

specific levels of specialization. Here is a task consisting of light, 
manual, semi-skilled work — assembling small parts of a typewriter or 
adding machine. A certain mechanism consists of two main subdi- 
visions, each made up of six individual parts. The two subdivisions 
are assembled separately, then joined together to form the completed 
mechanism. At present each worker performs the whole task, as- 
sembling one subdivision, then the other, and next completing the 
final assembly of the mechanism. 

Since there are a number of workers performing this same task, 
it is proposed that they be specialized, with one worker assembling 
only subdivision A, another assembling subdivision B, and a third 
worker completing the mechanism. The problem before us is how 
we may determine the relative efficiency of the newly proposed method 
as compared with the former procedure. 

The practical industrial supervisor, and probably most readers, 
would feel that they would want to see the job in operation, to know 
what the parts are like, what tools are used, and all of the details of 
the job. Naturally, all details should be before us if we are to work 
out the answer to our question. But we must stress the fact that we 
cannot answer our question adequately by applying "common sense" 
to the knowledge of the nature of the work. To attempt to do so 
would scarcely be sound scientific procedure, although it might be 
intelligent guessing. Yet, in practice, many problems of this sort 
are solved merely by inspecting the job and stating, ''Obviously, 
specialization is better for this job than the present method." 

We shall deliberately refrain, therefore, from giving any more 
detailed information about the nature of the work involved in our 
problem. We do not need these details in order to consider the 
general methodology which would be applied to a sound solution of 
the problem. No matter how familiar you are with the detailed oper- 
ations of the job, there is no equivalent substitute for an experimental 
study of both methods which would determine, in as direct a manner 
as possible, the relative efficiency of the two methods. 

For this kind of study it would be necessary that trials of both 
methods be carried on by workers of equivalent skill, with the same 
incentives, over a considerable period of time. This has frequently 
been done, but in most cases the experimenters have been satisfied 
to determine the relative productivity of the two methods. Suppose 
we find that the unspecialized method produces an average of 125 
pieces per day, while the specialized method yields 150 pieces per day, 
and that the difference is statistically significant. What have we ac- 
complished so far ? 

Ch. 2] 



Such a result is important, but it is only half the story. We know 
that the numerator of the fraction defining efficiency has increased, 
but what of the denominator? Until we know that, we have no way 
of determining whether the efficiency of the specialized method is 
higher or lower than that for the former method. To assume that 
the specialized method is ''better" is to ignore the denominator com- 

What must be considered in order to complete the analysis of the 
relative economy of the two methods of performing the task? In 
common terminology, we wish to know whether the individual worker 
is exerting himself more under the new regime, whether he is more 
fatigued at the end of the day, whether there is any accumulation of 
fatigue over a period of days or weeks, whether the new method 
makes him ''jittery" and has a harmful effect upon his appetite, his 
enjoyment of leisure, and his general health. As we have pointed 
out above, to consider these factors is not a matter of pure altruism 
on the part of management. Increases in these factors lower the 
"morale" of the workers, make supervision more difficult, increase 
the turnover and absentee rates, and generally affect the overhead of 
operation. Unless these elements are taken into account, the balance 
sheet for the job gives an entirely erroneous impression of the econ- 
omy of the operation from the point of view of management, as well 
as of the individual performing the task. 

It is possible that these potential losses are partially offset by 
certain gains on the part of the worker. He may get increased satis- 
faction from the work if his rate of output is higher. His increased 
pay may be important enough to him to compensate for the additional 
cost. Satisfaction from the work may be considered positively as an 
addition to the output beyond the number of assemblies produced, or 
it may be considered negatively as dissatisfaction and added to the 
cost. For reasons which will come out later, we shall place satisfac- 
tion and interest in the numerator of our efficiency ratio in place of 
adding dissatisfaction to the total cost. Numerically the result is the 
same in either case. 

These, then, are the general elements which must be considered 
along with production rate to form an over-all picture of the relative 
efficiency of the two methods under consideration. A statement of 
general terms is not enough, however, to permit us to compare effi- 
ciencies of different methods or conditions. We must be able to 
measure these changing factors in cost. The output in our example 
has increased 20 per cent as a result of the change of method. If fa- 
tigue, loss of health, or effort has increased by only 10 per cent, the 



[Ch. 2 

new method is more efficient than the old. The efficiency is not 20 
per cent greater, but at least it has increased, and the employer is in- 
terested in the additional output. Suppose, however, that the cost of 
the work has increased 20 per cent or more. Then it is doubtful 
whether the employer would gain in the long run by the use of the 
new method, even though the productivity has increased. With this 
increased cost of work, it is even doubtful that the productivity 
would remain at this high level over a long period of time. 

In the preceding paragraphs we have talked glibly and rather 
loosely about what should be done to analyze the efficiency of a method 
or condition of work. By our use of such easily understood words as 
"fatigue," "health," and "satisfaction," we may have emphasized the 
complete force of the term efficiency at the expense of making the 
problem appear simple. If this is the case, a second thought would 
show that the accurate study of efficiency is an extensive and difficult 
task. The first necessary step is fundamental research devoted to de- 
veloping methods of measuring these factors in the cost of w^ork. 
Only after some progress of this kind is made will it be possible to 
analyze efficiency in practical problems of the kind we have taken for 
our present illustrative example. 

Our first task in succeeding chapters of this book will be to evaluate 
progress which has been made along this general line of research, to 
consider what techniques are of some use in practical research prob- 
lems, and what further fundamental research needs to be carried out 
in order to increase progress in this field. 

It is the difficulty of the task of evaluating the cost of work which 
is the principal justification for considering only output in comparing 
methods of doing work in industry. Where this must be done, how- 
ever, it is dangerous to confuse the issue by equating measurement of 
output with measurement of efficiency. If the industrial manager 
was unable or unwilling to measure relative efficiency, he should say 
so, and speak of the relative productivity instead. 

Rate of Work, Efficiency in Emergencies, and the Aims of the 
Study of Efficiency. — Rate of work is obviously one of the factors 
which affects efficiency of performance. In many activities it has 
been found that a moderate pace is more economical than either a 
very slow or a very rapid speed of work. If the individual works 
very slowly, his "overhead" is high in proportion to what he accom- 
plishes, since he must go on living, standing, walking about. Also, he 
may use more awkward rhythms of movement. If his speed of work 
approaches the maximum rate of which the worker is capable, he ap- 

Ch. 2] 



parently adopts again relatively wasteful methods of work, just as an 
industry is likely to waste more in periods of peak production. Thus, 
output does not go up in proportion to the effort expended. 

If this relationship between rate of work and efficiency can be re- 
garded as a generalization (as it apparently can), it raises a question 
about the importance of the study of efficiency under emergency con- 
ditions. Those who read a discussion of efficiency of the individual 
worker during wartime were likely to feel that this is a topic worth 
considering in peacetime, but that it had little relevance to wartime 
emergencies. One cannot achieve maximal efficiency at high speeds 
of work; therefore they concluded that we must await the more 
moderate pace of peacetime before taking an interest in the problem. 

The determination of optimal efficiency is always, however, a rela- 
tive matter. Under any given conditions, certain factors cannot be 
changed. Our problem is to adjust to the most satisfactory level 
those factors which are under control. During war we do not have 
complete liberty to choose the rate of work which leads to maximum 
output per unit of cost. The problem is, rather, to determine con- 
ditions of efficiency which can be controlled and which improve ef- 
ficiency at a high rate of performance. Variations in speed still need 
to be considered, since the speed could increase to the point where 
errors, accidents, and disruptions in production would be so frequent 
that the net output would be reduced. 

Another way of looking at this same problem might also be 
enlightening, even though it is largely speculative at present. In 
emergency conditions, the worker is willing to devote more effort to 
his work than he is in normal times. Nevertheless there are limits 
to the total input which is available, especially if the "emergency" is 
long lasting. Under these conditions, the problem of efficiency is to 
find out how this limited input can be used to greatest advantage. 
Rate of work is a variable within certain limits because the worker 
might use his quota of input in six hours per day or spread it over a 
longer working day. The only limitations upon rate of work are that 
the worker is expected to use his quota completely, and his rate of 
expenditure must not be so great as to introduce errors and accidents. 
In other words, rate of work is still a variable factor in efficiency, 
even under these more limited conditions where we cannot aim for 
the absolutely optimum speed of performance. 

It is also conceivable that the optimal rate of performance might 
not be put into practice even in peacetime, for a number of reasons. 
The optimal rate for a single individual might be so slow that it would 
produce a labor shortage and an inadequate supply of goods. Even 



[Ch. 2 

from the individual's point of view, the optimal rate might not be 
satisfactory. By optimal rate we mean that rate at which there is a 
maximum output per unit of input. Suppose, however, that this leads 
to a very low total cost to the individual per day. Perhaps even after 
we have allowed for the effort he can expend upon his hobbies, 
amusements, home affairs, and the like, the worker is still left with 
a considerable reserve which is not used at all. 

The problem of rate of work is, then, very complex at all times. 
The discussion in the preceding paragraphs was a statement of vari- 
ous possible relationships rather than of established facts. The mere 
existence of these possibilities should indicate that a more careful 
analysis of over-all efficiency is needed, both in wartime and during 
peace. Conjectures and opinions are made the basis of many argu- 
ments about problems of fair wages and hours of work, restriction 
of output, "featherbedding," and the like. These arguments must 
remain in the realm of conjecture until more adequate research upon 
efficiency can be related to these problems. 

Applications of the results of research will not always be directed 
toward absolutely optimal conditions for all factors. There would 
be no point in allowing effort to go to waste in achieving optimal 
efficiency. Other factors, however, will not only raise efficiency at any 
given rate of performance, but they will also permit a higher output 
for any given expenditure per day. Whether we seek an optimum 
condition, or a condition which merely avoids extreme iVzefficiency, 
the general concept of efficiency is fundamental in our thinking. 

Poffenberger has stated an ''ideal of human efficiency" which 
"would be the production of the maximal output of the highest 
quality in the shortest time, with the least expenditure of energ)^ and 
with the maximum satisfaction." ^ This statement is, in the opinion 
of the present writer, incomplete in several ways. Only energy and 
time are considered as elements of input, although Poffenberger 
shows in the remainder of his discussion that he intends to include 
fatigue and other elements of cost as well. Apart from this conden- 
sation of the statement, there is a more serious drawback. A maxi- 
mum of output and a minimum of cost are incompatible. Instead, 
we must seek a compromise between these two distinct aims. A better 
statement might be : "the greatest output and satisfaction which can 
be attained without excessive cost, understanding by excessive cost a 
level of effort which leads to accumulation of fatigue, inability to 

2 A. T. Poffenberger, Principles of Applied Psychology, New York, D. Apple- 
ton-Century Co., Inc., 1942, p. 364. 

Ch. 2] 



enjoy leisure, loss of health, premature aging, increased liability to 

This revised statement of the broad aims of research upon effi- 
ciency should, we believe, answer a critic of Poffenberger who raises 
the question ''Why be efficient?" ^ This critic, Johnson, believes that 
many of us are too efficient, that those who grow fat might be better 
off to use more energy in their work, rather than less. Johnson's 
notion of efficiency is restricted to mechanical efficiency; it ignores 
the possibility that a performance which has a low cost in terms of 
energy may have a high cost in terms of other elements included in 
our own broader meaning of efficiency. Even if we restrict atten- 
tion to the mechanical side of efficiency, the choice of a method of 
work need not be dictated solely by an optimal efficiency ratio, as the 
preceding discussions should indicate.* 

Total Efficiency. There is another stumbling block in the 
measurement of total cost of work beyond that of the number and 
complexity of factors included. This difficulty is that of summing up 
the elements of cost. Obviously the cost of human work to the in- 
dividual cannot be placed on a dollars and cents basis, even very 
roughly. It is equally evident that many of the elements, such as 
discontent or loss of health, cannot be expressed in energy terms. 
How, then, is it possible to make any useful research approach to the 
problem of efficiency? It can only be done piecemeal and indirectly. 
If we are able to hold most items of cost at a nearly constant level 
while one item varies over a wide range, we can draw conclusions 
about efficiency from this one factor alone. If a given factor reduces 
all or most elements of cost at the same time, again it is easy to draw 
our conclusions. 

When, however, we find divergent effects upon the various items 
of cost, the problem is one which can be solved only in terms of some 
theory of value. Suppose, for example, that a change in working 
conditions increased fatigue but decreased discontent. The decision 
as to the effect upon efficiency or the desirability of the conditions can 
be made only by coming to a decision about the relative importance of 
fatigue and discontent to the individual, and in the long run. Perhaps 
such a question could eventually be answered by research. 

3 See H. M. Johnson's review of Poffenberger's Principles of Applied Psychol- 
ogy, American Journal of Psychology, 1943, 56, 613-619. 

* Parenthetically, we might also raise the question of whether growing fat is 
necessarily an indicator of high efficiency, even of the mechanical kind. Perhaps 
it indicates the efficiency of the digestive and glandular processes more exactly 
than it does the efficiency of the activities involved in work. 



[Ch. 2 

This is an instance, like others which we shall meet later, where 
the applied psychologist has to take a somewhat different approach 
from the psychologist who is interested in ''pure science." The latter, 
particularly one who calls himself an ''operationist," would say that 
the problem of efficiency, as we have formulated it, is not a scientific 
problem. What do we gain, he would say, by using a term which 
refers to elements which cannot be measured or satisfactorily com- 
bined? He would prefer to take one aspect of cost which is readily 
open to observation and measurement, and define the cost of work in 
these terms. Then, when he uses the term efficiency, he will refer 
to a perfectly definite, measurable thing. Suppose, for example, that 
we define the cost of work in terms of the oxygen consumed per unit 
of work, which is the common way of measuring the mechanical ef- 
ficiency of muscular work. We would decide that the term efficiefuy 
will mean nothing more nor less than the oxygen consumpion per 
unit of work. 

The trouble with this approach, from the point of view of the ap- 
plied psychologist, is that it applies only to work where physical 
energy expenditure is great. In other tasks, other elements of cost 
are more important, and we cannot diminish their importance by re- 
fusing to consider them. It is better, therefore, frankly to face the 
fact that some of the elements of cost which would be essential to a 
complete solution of the problem of efficiency are not amenable to 
available methods. It makes us realize the limitations of conclusions 
based upon the measurable elements, and shows us the direction which 
research must take in order to reach more complete solutions of the 

Interrelationships of Factors in EfHciency The various factors 

in efficiency are interlocking in a variety of ways. Some of the inter- 
relationships have been discussed at various stages of the development 
in this chapter. It might be well, however, to bring together all these 
interdependencies to form an integrated pattern, and to bring out 
more clearly the meaning of the various terms which we have used in 
discussing efficiency. 

A worker is performing a task under specified conditions and with 
a specified method. At a given moment, and under these specified 
conditions, the worker has an upper limit of speed of work beyond 
which he cannot go. It is this momentary, conditional limit which we 
shall speak of as his capacity.^ Effort is a term describing the rate of 

5 It will be noted that this notion of capacity is distinct from the meaning usually 
implied in psychological texts. Usually, capacity is used interchangeably with 
aptitude, or potentiality to learn a given performance. In the context of efficiency. 

Ch. 2] 



work in relation to capacity. This measure of effort can thus be 
changed by either a change in'rate of work or a change in capacity. 
If, for example, the capacity of the worker declines while his absolute 
rate of work remains constant, his effort has increased. Conversely, 
rate of work can increase without any increase in effort. If a new 
method of doing the job is introduced, a method which makes the 
work easier and raises the capacity of the worker, the rate of work 
can be increased proportionately, to leave the level of effort at its 
original value. 

Other phases of the cost of work are, in turn, functions of effort. 
Thus we shall assume that fatigue, energy expenditure, loss of health 
are each dependent upon the degree of effort involved in an activity. 
The exact relationships are not known. It is likely that a different 
function describes the relationship of each of these factors to effort. 
Each is probably also dependent upon other things besides effort in 
the task itself, but our main concern is with the relationship to effort. 
It is also very probable that the functions are not linear. That is, 
fatigue is not proportional to effort, but probably increases dispropor- 
tionately when effort approaches the maximum. Fatigue may also 
increase when effort goes down to a low level. It is also likely that 
these relationships vary with the kind of work. Energy expenditure, 
for example, may bear a close relationship to effort in heavy muscular 
endeavors, but show little effect of other kinds of effort. 

It is apparent that a single research project cannot take all of 
these potential relationships into account. Ordinarily it is necessary 
to limit the study to the observation of a few factors. Thus, energy 
expenditure may be the only element of cost which can be accurately 
observed; in other instances attention must be concentrated upon 
fatigue. It is seldom that effort is taken into account directly and 
related to fatigue or other elements of cost. Instead, fatigue is re- 
lated directly to some external factor such as noise, or to rate of 
work as a substitute for a measure of effort. 

In spite of these necessary shortcomings in research procedures, 
all the interrelationships just discussed should be kept in mind in 
interpreting the results of specific experiments. Otherwise there is 
a serious possibility of error or misconception of the meaning of the 

however, the meaning given in this text is usually adopted, separating the concepts 
of capacity and of aptitude. Capacity is thus a limit v^hich is never reached in 
practice, and is a concept closely similar to the concept of limit in mathematics. 
Even though this limit is never reached, and therefore capacity cannot be directly 
measured, it can be estimated indirectly, and is a construct which is useful in our 
thinking about human activity, just as the concept of limit is useful in mathematics. 



[Ch. 2 

Summary of the Concept of Efficiency. — The term efficiency in- 
volves many complex factors. In this book it will designate a ratio 
between output and input (cost of work or effort). To be complete, 
the output term should include not only the "objective," tangible 
products of the work, but also the pleasure and satisfaction which 
the worker may derive from his accomplishment. The input factor 
of the ratio involves a complex set of factors, since we wish to include 
all elements of the cost of work to the worker. 

It would be pleasant to simplify the concept of efficiency by re- 
ducing the number of terms which are to be taken into consideration, 
and by simplifying the elements themselves. Such a simplification 
would be a loss to applied psychology, however, since this is a field in 
which we must accept our problems from everyday living, and do our 
best to cope with them as they appear. We cannot avoid these prob- 
lems forever, even though we may not feel able to cope with them in 
their complete form at the moment. To retain the complex meaning 
of the term efficiency is to lay out a general program for applied psy- 
chology which is to be solved little by little. The purpose of the 
coming chapters will be to survey what has been done in this program. 
To ignore the parts of the program which are still incomplete, would, 
however, give the reader an utterly false notion of progress in ap- 
plied psychology. 

There can be no doubt that progress in this program will have 
marked practical effects. Some of our most important problems of 
personnel and labor relations, such as fair wage rates, hours of work, 
work methods, morale, and output restriction, depend in their last 
analysis upon the facts of efficiency for their solution. 

In this discussion of the implications of the term efficiency, we 
have, perhaps, stressed the study of input too much in relation to the 
study of output. The reason for this is that we wished to emphasize 
those phases of the problem which are most commonly neglected or 
minimized. We do not intend, however, to go to the other extreme 
and to neglect the very important study of productivity. Output is 
an important subject for study in its own right, as well as an integral 
part of the study of efficiency. If the effect of a certain factor upon 
output, but not upon efficiency, is known, this information can be 
useful and we shall not intentionally slight it. 

Chapter 3 


The measurement of output is relatively easy — at least it is easy 
enough so that we can take that phase of efficiency measurement for 
granted until we have become more familiar with the problems which 
we face in dealing with the cost of work. The purpose of this chap- 
ter, and of the next, will be to familiarize the reader with the general 
techniques for studying and evaluating the cost or input. We shall 
consider the techniques of measuring the cost of work according to 
the class of activity for which each technique finds its principal use. 

It is common to divide work into two main classes, "mental" and 
''muscular" (or "physical"). Such a separation originally implied a 
separation of mind and body which scientists no longer regard as a 
valid or useful separation. Research studies of the cost of work still 
fall roughly into two general classes, however, even though the orig- 
inal basis of the distinction between "mental" and "muscular" work 
has been lost, and the adjectives no longer have their original 

We shall include under muscular work only those kinds of activity 
in which the primary aim is the development of mechanical force. 
Muscular work would include tasks which make heavy demands upon 
the whole muscular system — running, walking, carrying large loads, 
shoveling, scrubbing, and the like. It would also include those tasks 
which involve the development of force by a restricted muscle group, 
such as lifting a weight with one arm or one finger while the rest 
of the body is supported. 

Perhaps it would be better to speak of "sedentary" work rather 
than "mental" work as our second main division of work. The 
reason for preferring such a terminology would be to avoid the un- 
fortunate implication that "mental" work involves only the "mind," 
which in popular usage frequently implies the brain alone. Neverthe- 
less we shall, for the sake of brevity, use the term "mental" with the 
warning that it means only that we are dealing with a kind of work in 
which control, timing, skill, and direction of activity are more im- 




[Ch. 3 

portant than muscular force, whether or not muscular activity is in- 
volved. Since most occupations of the sedentary kind fall into this 
classification, it seems that ''sedentary" would be as good an adjective 
as ''mental" although it is a less familiar usage. 

There are many activities which are difficult to classify into either 
of these main divisions. All would agree that loading a wagon with 
hay belongs to the muscular class of work. Solving a problem in the 
calculus evidently belongs to the sedentary or mental classification. 
What about such a task as typing, however? Such a task involves 
some of the characteristics of each class. Muscular force is respon- 
sible for the impression on the page, but the differentiation between 
adequate and inadequate performance is primarily in terms of ac- 
curacy and pattern of movements rather than force. For some 
purposes, therefore, we could study typing as a kind of muscular 
work, but we should be considering only one aspect of the task. In 
the over-all view of the task, typing should probably be classed with 
mental work. This example should emphasize the fact that the mean- 
ing of muscular work, as used in this discussion, is more restricted 
than it is in common usage. Some tasks which we think of as manual 
jobs may still not be called muscular work because force is not a 
primary feature of the task. 

A classification of kinds of work should probably be made eventu- 
ally on the basis of the internal bodily activities involved, and in terms 
of the kind of effect which that work develops in the body. The pres- 
ent classification is partly in terms of the nature of the outcome of the 
work, as well as conjecture as to the bodily activities. We need some 
kind of preliminary classification, however, until a more adequate one 
can be devised ; otherwise it is difficult to organize the material upon 
the cost of work now available. 

In the following discussion of muscular work, we shall find that 
several special divisions of the class are beginning to emerge. The 
effects of fatigue in these classes appear to be quite distinct. It will 
therefore be useful to keep these subdivisions in view as we go along. 
The divisions are listed in Table 2. 

Much more is known of the bodily effects of some of these types of 
work than of others. For example, more research has been directed 
toward the first type of "dromal" task than to the steady grind, prob- 
ably because the first lends itself more readily to laboratory investiga- 
tion, although the second is of greater practical significance. 

The central interest of this book is not in muscular work, although 
the present chapter is primarily concerned with the topic. Our pur- 
pose is to introduce the reader to the techniques of measuring cost of 

Ch. 3] 



work which have been developed in this field, and to do this without 
presupposing much background in physiology. This survey is essen- 
tial background for studying the cost of sedentary or mental work. 
As psychologists, we are more concerned with mental work, and it 
occupies a more important place in business, industry, and education. 
Nevertheless, research upon muscular work has an interest and im- 
portance of its own, as well as furnishing a needed background for 
other studies. 

Types of Muscular Work * 

1. Rapid expenditure of a large amount of energy ("dromal" tasks). E.g., 

running a competitive race. 

2. Steady grind. E.g., a long day, walking under a load, stoking-, etc. 

3. Repetitive local task v^ith high local energy expenditure. E.g., lifting a 

weight with the finger. 

4. Postural restrictions ("static work"). E.g., maintaining an awkward 

position ; holding a weight in outstretched hand. 

* Modified from T. A. Ryan, Varieties of fatigue, American Journal of Psychology, 1944, 
57, 565-569. 

Energy Cost of Work (Fuel Consumption) 

In discussing muscular force, we naturally consider first its cost 
in terms of the energy (in calories or B.T.U.) released in the body — 
energy which is ultimately derived from food. Because the body has 
so many provisions for storage of energy-producing materials, food 
intake can give us only a very rough indication of the energy used at 
any given time, even after we allow for waste through excretion. 
Oxygen consumption during any interval of time is, however, propor- 
tional to the energy released in the body, since oxygen cannot be 
stored in quantities large enough to affect the results. To be sure, the 
energy equivalent of a given volume of oxygen varies somewhat with 
the nature of the food which is being oxidized. It is possible to esti- 
mate the proportion of various nutrients being used and make allow- 
ances for this variation, but with an average diet the variation is not 
great enough to affect the results in any important way. One liter of 
oxygen corresponds to about 5 (large) calories (Cal.) of energy lib- 
erated (4.686 Cal. for pure fat to 5.047 Cal. for pure carbohydrate).^ 

1 F. G. Benedict and H. Murschhauser, Energy transformations during hori- 
zontal walking, Carnegie Institution of Washington, 1915, Publication No. 231, 
p. 8. 



[Ch. 3 

This correspondence between oxygen consumption and the energy 
released in the body holds over a period of time, but it is inexact for 
any short span of time, such as a few minutes. If a man engages in 
heavy muscular effort for a short time, we find that his oxygen con- 
sumption is high not only during the time when he is working, but 
also for a considerable period after he has stopped work altogether. 
During work, he contracts an oxygen debt which is repaid during the 
following rest period. 

The explanation for the appearance of the oxygen debt is that the 
oxygen is not used directly in the chemical processes of muscular con- 
traction themselves. It is known that these processes can go on in 
the absence of oxygen. The muscles soon lose their power to con- 
tract, however, unless oxygen is available. Thus the oxygen is neces- 
sary for rebuilding the contractility of the muscle, but it is not 
necessary that all this recovery process be accomplished during the 
work itself. In fact, experiments indicate that a certain amount of 
lag is beneficial, and that the muscles operate more effectively after a 
certain oxygen debt has been built up. 

In order to determine the total energy cost of a certain muscular 
task, therefore, it is necessary to find the total amount of oxygen 
which is required to bring the muscles back to their original state 
before the work began. This means that the oxygen debt must be 
included in the total oxygen consumption. We must measure the 
extra oxygen used during work, and add to that the extra oxygen 
consumed during the rest period after work. It is especially im- 
portant to take the oxygen debt into account for short sprints or other 
activities involving a rapid expenditure of energy. 

In studies of work, such as lifting weights, the energy involved in 
doing the work is separated from the energy required to maintain the 
body in its resting state. Suppose, for example, that a man uses an 
average of .25 liter of oxygen per minute while he is resting. While 
he is lifting a given weight in a prescribed manner, he averages .45 
liter of oxygen per minute, (including the ''oxygen debt"). The dif- 
ference represents the amount which is required to supply energy for 
the work. Multiplying the difference by 5 gives us the energy in 
calories which is required to perform the task for one minute, (in this 
example, 1 Cal). (See summary of computations in Table 3 on 
page 37.) 

Let us suppose that the work consisted in lifting a 25-pound weight, 
at the rate of four lifts per minute through a distance of three feet. 
He has thus accomplished 300 foot-pounds of work, and has used one 
calorie in doing so. If we had a perfect machine in which all energy 



went into useful work, the amount of energy this man has consumed 
would have been converted into 3085 foot-pounds of work. There- 
fore the mechanical efficiency of this performance is about 10 per 


Rough Computation of Mechanical Efficiency 

Input : 

Total oxygen used during work 

(including oxygen debt) 45 liters per min. 

Oxygen consumption during rest 25 " " " 

Net oxygen consumption 20 " " 

(increase due to work) 
Conversion factor X 5 

(1 liter = 5 Cal.) 
Net input 

(energy consumption due to work) 1.00 Cal. " " 

Work equivalent of input 3085 F. P. 

(1 Cal.= 3085 F. P.) 

(external work done) 300 " " per min. 

w Physical Work 

Mechanical efficiency = 100 X — — ; — — 

(of muscular system used) equivalent of input 

= 100 X = 10 per cent 

The way in which this method could be used for determining the 
value of different methods of work is obvious. Suppose, for ex- 
ample, it is decided to double the load and cut the rate in half. Then 
the work per minute would be the same as before. If, now, it is found 
that the oxygen consumption is more than it was before, the new 
procedure is clearly less efficient. 

In laboratory studies it is easy to choose tasks which permit ready 
computation of the work done in standard physical units such as 
foot-pounds. The results of these experiments are suggestive and 
useful in many ways, but their application to different tasks with 
different patterns of muscular movement is very doubtful. Fortu- 
nately, many tasks can be studied by the metabolic method, even 
though their output cannot be expressed in a standard set of units. 
Horizontal walking, for example, could be converted into equivalent 
work units only by a complex physical study of the forces involved. 



[Ch. 3 

It so happens that this difficulty is not a very serious one. We very 
seldom desire to compare the efficiency of performance of two dis- 
similar tasks, such as lifting weights and walking. If we are com- 
paring different conditions of walking, or different rates of walking, 
we can simply use any units of output which are convenient — distance 
travelled, or distance times the weight of the body and its load. We 
are not able to speak of "per cent efficiency" in these circumstances, 
since input and output are not in comparable physical units. We se- 
cure an index proportional to the efficiency, however, and we are 
interested only in relative efficiency anyway. In some cases we can 
hold the output constant and observe the metabolic cost alone as the 
indicator of efficiency. 

Applications to Muscular Work. — Table 4 shows the results of 
one study of efficiency in a muscular task in which metabolic measures 
are employed. Atzler shows here the relation of height of a crank 
axis and the load to energy consumption per unit of work. In other 
phases of his experiment he also studied the effect of the rate of 
turning and the radius of the crank. The latter factors were held 
constant in the experiments which are reported in Table 4. 


Cost of Work in Lifting Loads by Crank * 
(Calories per meter-kilogram) 

Height of 

Load (meter-kilograms 

per turn) 

Crank Axis in cm. 






























* E. Atzler, Probleme und Aufgaben der Arbeitsphysiologie, Ergebttisse der Physiologk, 
1928, 27, 720. This article (pp. 709-779) gives a number of other examples of tiis same 
method. Other examples are to be found in another article by the same author, Arbeits- 
physiologie: Zweiter Teil, Ergebnisse der Physiologie, 1939, Ifl, 164-291. 

These results are a good instance of the manner in which experi- 
ments upon metabolic cost could be used in deciding questions of the 
design of machines and tools. It will be noted that greatest effi- 
ciency is obtained when the crank axis is 114 cm. from the floor, re- 
gardless of the load. That is, the energy (calories) per unit of work 
(meter-kilogram) is at a minimum for this particular height. If the 
height is increased or reduced, the efficiency is reduced. 

Ch. 3] 



It will also be seen that the very light load and the heaviest load 
are both less efficiently handled than the intermediate loads. The op- 
timum load varied, however, with changes in the height of the axis. 


Relative Energy Cost of Pulling vs. Pushing* 
(Calories per meter-kilogram) 

Weight Pulling Pushing 

(Two hands, grip 100 cm. (Crip 75 cm. from 

from ground) ground) 
10.27 10.67 Cal./mkg. 9.16 
11.64 10.21 9.04 
13.56 9.76 8.72 
16.06 9.90 8.96 

* Atzler, op. cit. supra, p. 742. 

This set of data also illustrates a common inadequacy of experi- 
ments upon energy cost. This is the tendency to ignore individual 
variation and chance variation, which is understandable when we con- 
sider the great number of observations which must be made on each 
subject. Atzler's study involved only two subjects. They were, how- 
ever, quite different in build and weight, and their results were closely 
similar. For this reason Atzler feels that a study of a wider range 
of individuals is unnecessary. Perhaps this is justified in this case, 
but the same tendency to work with very few subjects is often found 
in studies where there is much more likely to be significant variation 
from individual to individual. 

We are not told how many observations were made to establish 
each of the values of the table (another common tendency among 
some physiological investigators). For this reason we cannot say 
whether some of the slight variations reported there are more than 
might be expected by chance. 

Another kind of work is analyzed in Table 5, also by Atzler. This 
table represents the cost of work done in drawing or pushing carts 
with varying loads. It will be seen that pushing is more efficient 
than pulling the load for each of the weights studied. It will also be 
noted that, within the range of this experiment, the efficiency in- 
creases as the weight of the load increases. In another part of the 
study it was found that the handgrip for pulling the load was most 
efficiently placed at 100 cm. from the ground. A variation of 15 cm. 
in either direction increased the energy consumption per unit of work. 



[Ch. 3 

A third example is based upon a large number of studies of walk- 
ing and running. Table 6 shows results assembled by Poffenberger 
to indicate the efficiency of varied rates of locomotion. 

TABLE 6 * 
Energy Expended in Human Locomotion 


Lying down in post-absorptive state 

Standing still 

c.c. of O2 
per Minute 

C.C. of O2 
per Mile 

Walking 1.66 m.p.h 



2.00 m.p.h 



3.00 m.p.h 



4.00 m.p.h 



5.00 m.p.h 



Running 6.50 m.p.h 



7.50 m.p.h 



9.00 m.p.h 



10.00 m.p.h 



* Reprinted from A. T. Poffenberger, Principles of 
D. Appleton-Century Co., Inc., Copyright, 1942, p. 117. 

Applied Psychology, 

New York, 

This example brings up the question of how much of the oxygen 
consumption to include in the cost of work. The *'c.c. of O2 per mile" 
listed in Poffenberger's table is based upon the total oxygen consump- 
tion, including not only the oxygen employed in the muscular activi- 
ties of locomotion, but also oxygen which is used for the basal 
metabolism. He is concerned, therefore, with something more than 
we should include in the strict meaning of ''mechanical efficiency.' 
Some of the overhead is included as well, in the sense that oxygen 
used in general maintenance of the body is also counted. If we were 
dealing with different methods of doing a job at the same fixed rate, 
overhead would remain constant and it would make little difference 
whether it were included or not. When the rate of work varies, the 
overhead per unit of work (per mile, in this instance) varies. That 
is, the 237 c.c. resting consumption is used sixty times in a mile at the 
rate of one mile per hour, but only six times in a mile at the ten-mile 

The problem of whether or not to include the overhead is to be 
decided by the purpose of a study. If a physiologist wishes to find 
the mechanical efficiency of a set of muscles used in a certain task, he 
should use only the excess of oxygen over resting values in computing 
input. If we are interested in efficiency of the total body, and if rate 


of work is not a variable factor in our experiments, we do not need 
to eliminate the resting metabolism from the total input. 

If we are interested in efficiency of the whole body, but if rate of 
work is a variable factor, the problem becomes more complex. An 
argument in favor of subtracting the resting value would be that the 
man must go on living anyway, and that the work is not directly re- 
sponsible for this cost. In the case of a real job, the "overhead" 
should be determined separately, and the basal metabolism for twenty- 
four hours would be included in a complete analysis of the input. If 
a man runs ten miles per hour, the basal metabolism per mile during 
the running is much lower than it is when he is walking at one mile 
per hour. If he runs ten miles per hour in making a certain trip, 
however, he must rest more frequently and longer. His basal re- 
quirements go on during those rests as well as during the time of 
running itself. If he isn't able to go as far in a day when he runs as 
he goes in walking at a slow pace, the basal metabolism per mile would 
be greater at the faster pace. 

To carry further the interpretation of the problem of efficiency 
of walking, it would, therefore, be necessary to know more of the 
conditions under which the activity is to take place — how far the sub- 
ject is to walk, how his rest is to be distributed, and so on. The 
strictly mechanical efficiency of walking cannot be computed from 
the data of Poffenberger's table because the various values were not 
taken from the same subjects. If we did make such a computation 
from the above table, we should find that the most efficient rate is 
slower than the four-mile-per-hour optimum indicated by the values 
of the last column. In fact, if the data were all from the same sub- 
jects, the optimum would be below the slowest rate shown in the table. 

Metabolism in Mental Work. — Tasks like mental calculation or 
memorizing, which can be carried on with a minimum of muscular 
activity, cost the organism very little in terms of energy expenditure. 
The estimates vary with the conditions of the experiment, but most 
of them are low.^ We shall not go into the theoretical implications of 
this finding here. It suffices to point out that oxygen consumption is 

2 F. G. Benedict and C. G. Benedict, Mental effort in relation to gaseous 
exchange, heart rate, and mechanics of respiration, Carnegie Institution of Wash- 
ington, 1933, Publication No. 446. D. E. Rosenblum, Untersuchungen iiber den 
respiratorischen Gasstoffwechsel und Energieverbrauch by geistiger Arbeit, Arbeits- 
physiologie, 1932-1933, 6, 214-234. See also G. H. Rounds, H. J. P. Schubert, and 
A. T. Poffenberger, The effects of practice upon the metabolic cost of mental work, 
Journal of General Psychology, 1932, 7, 65-79. The latter authors give a larger 
estimate than Benedict because they do not try to minimize the motor involvement 
of the mental tasks. 



[Ch. 3 

not a reliable measure of the cost of work in tasks which involve 
muscle activity only incidentally. 

Bills ^ has argued that even though the caloric output of the body 
is increased only slightly by these sedentary activities, the increased 
energy expenditure may involve such a small area of the nervous 
system that it really represents intense activity at that point. He 
draws an analogy with an electric clock which uses little electric cur- 
rent, although the current is essential to its functioning. All this 
may be true, but the fact still remains that we cannot detect these 
changes in the total calorie output of the body except under very 
carefully controlled and unnatural conditions, and even then not very 
reliably. Until we know where these foci of energy release are lo- 
cated in the body, and have methods of measuring these local events, 
metabolic studies will be of little value in the study of efficient con- 
ditions of mental performance. 

Barmack's ^ work supports the position w^e have taken here. While 
he finds that there is usually an increased oxygen intake at the start 
of a new task, what happens after that is exceedingly variable. Bar- 
mack believes that there is a tendency for metabolism to fall when the 
subject becomes bored with the work. This leads to a paradoxical 
situation. We should expect that the effort required to continue 
working would increase when the subject is bored and disinclined to 
work. If so, we should expect oxygen consumption to increase if it 
bears any relation at all to the cost of this kind of work. 


From our common observation we would expect that fatigue 
would be one of the important factors to be taken into account in 
evaluating the cost of work. The worker on a job for which he is not 
fitted may grow so tired that he is willing to take another at lower 
pay. He does not feel that his present job pays him for the extra 
input which it demands. Especially in those types of work in which 
metabolic indices cannot be applied, it would seem that a measure of 
fatigue should be applied. 

Our common observation is probably correct, but it is not an easy 
suggestion to carry out. One of the primary difficulties is that the 
term "fatigue" is not clearly enough defined to aid us in a search for 
a measure. Not only is the term vague as it appears in common 
usage, but it has been difficult for scientists to agree upon a common 

s A. G. Bills, The Psychology of Efficiency, New York, Harper & Bros., 1943. 
* J. E. Barmack, Boredom and other factors in the physiology of mental effort, 
Archives of Psychology, 1937, No. 218. 

Ch. 3] 



meaning of the term which would put their research upon a straight- 
forward basis. 

Unfortunately, fatigue refers to several distinct phases of the cost 
of work. The several items included under this heading may be 
loosely correlated with one another, as implied by the use of the com- 
mon term ''fatigue" to describe them all. At the same time there is 
evidence that the correlation is not close enough to permit us to 
measure one kind of fatigue alone, and rest secure in the belief that 
we have accounted for all the fatigue cost by means of a single index. 

As a general statement, we may say that fatigue refers to a large 
number of residual effects of work. That is, it refers to any inhibi- 
tory effects of activity which carry over to the period after work 
ceases, and also to effects which cumulate during a period of con- 
tinuous work. The only residual effects which would be excluded 
from the definition would be permanent changes or non-reversible 
changes, such as structural damage to some tissue or organ (deaf- 
ness, lameness, heart disease, etc.) and those changes which repre- 
sent a facilitation of performance (learning, warming-up, positive 

Such a definition may be regarded as too broad to be useful. In 
fact, the term fatigue is useful only as a general designation for a 
broad area of research. The whole area is important, however, in any 
study of total human efficiency, since all the effects classed under the 
heading of fatigue are phases of the cost of work to the individual. 
It does not matter whether we decide to restrict the term to a single 
unitary phenomenon or not, the other factors also must be con- 
sidered, whether they are considered under the heading of fatigue or 
under some other designation. We do not believe it is possible to 
restrict the term in this manner, for several reasons. For one thing, 
common usage is against the restriction, and we should be inviting 
misunderstanding. Secondly, attempts to narrow the meaning of the 
term have not been very successful because it has been difficult to 
delimit the boundaries in an unequivocal fashion. 

This broad topic of fatigue covers certain important focal areas, 
each one of which is important, and each one of which is a very 
complex problem for investigation. The subdivisions of the field 
may be listed as follows : 

L Feelings of tiredness, weariness, or exhaustion (sometimes called 

''perceived fatigue"). 
2. A reduction in the worker's capacity to perform his job, as a 

result of previous performance of the same task. 



[Ch. 3 

3. Changes in other activities and capacities (including changes in 
physiological function and changes in ability to perform psycho- 
logical activities other than the job itself). 

We may think of all these phenomena as dependent upon certain 
fundamental bodily changes — changes in neural function, in the meta- 
bolic activities of certain organs, and so on. If it is preferred, we 
may consider that "true fatigue" consists of these fundamental pat- 
terns of bodily change, and that the three sets of phenomena described 
above are all indirect effects of the same set of ultimate factors in the 
body. The lack of precise correlation betw^een various effects, such 
as w^eariness and loss of capacity, may be explained as due to the 
different indicators reflecting different phases of the fundamental 
bodily changes. 

Even though much physiological research seems to reflect the 
above point of witw that ''true fatigue" is something which underlies 
these more superficial expressions of fatigue, it is not a point of view 
which is of much value for most practical studies of worker efficiency. 
Perhaps some day the situation will change and justify such a notion 
of "true fatigue." At present, however, it is only occasionally a use- 
ful concept. First of all, the items listed above as phases of fatigue 
are much closer to direct observability than the underlying changes in 
physiological function. Secondly, there is little evidence to support 
the point of view that there is any single pattern of organic change 
which is the characteristic basis of all the aspects of fatigue mentioned 
above. On the contrary, as we shall see, present evidence indicates 
that similar changes in capacity to perform may depend upon one set 
of bodily conditions under some circumstances and upon an entirely 
different set of conditions if the situation is changed. Even with this 
concept of fatigue, therefore, the term still designates a very broad 
class of phenomena. 

Feelings of Tiredness. — Although the principal experimental 
studies have been concerned primarily with the capacity changes, it 
should not be assumed that the feeling of tiredness is unimportant. 
Perhaps the layman is not very likely to make such an error, but the 
professional writings upon the subject of fatigue sometimes leave 
the impression that the feelings of the worker are somehow less im- 
portant than the more "objective" effects of work. Such an impli- 
cation is probably not intended; it may result from the difficulty of 
securing a workable research approach to the problem of feelings, and 
a satisfactory definition of what is meant by the common-sense term 


The reaction of the layman to this discussion is hkely to be some- 
thing Hke this : ''I don't see why a definition is so important, or even 
necessary. I understand what others mean when they say they are 
'tired out/ 'weary,' or 'exhausted.' Why not treat this as some- 
thing which is self-evident? We don't have to define what we mean 
by 'red' or the 'taste' of an orange, so why should it be necessary 

Such an objection has its points, but there are good reasons why 
a definition is needed, even so. First of all, perhaps we are all re- 
ferring to the same things by these terms. A scientific study of the 
problem requires something more, however. Suppose we are faced 
with the problem of which of two methods of doing a given job is 
more efficient. We might ask the workers to use each method and 
then ask them which is more fatiguing. Unless they work for a long 
time with each method they would not be able to assess them because 
the first adaptation to the method might be quite different from what 
happens after they are thoroughly accustomed to the conditions. 
Therefore we should have to require that our subjects compare de- 
grees of fatigue from memory over a considerable period of time. 
Even if this were possible, it would still be impossible to say how 
much one method excels the other. If the production rate on the two 
methods is changing at the same time, the problem is still more 
complicated. We should have to ask the subjects to estimate a 
quantity of fatigue so that we could estimate the amount of fatigue 
per unit of work. 

There have been some attempts to use a refinement of this pro- 
cedure, in the form of a rating scale. Poffenberger, for example, 
presents the curves of Figure 2 derived from a study of output and 
feelings of tiredness.^ The subjects rated their feelings at regular 
intervals during the experiment, on a scale ranging from + 3 to — 3. 

Notice, however, that these results are not used as a measure of 
quantity, but simply as a method of comparing relative values at dif- 
ferent times on the same day. And this is as far as the method could 
be carried. 

Fatigue Regarded as Reduced Capacity for Work. — In order to 
avoid the above difficulties with perceived fatigue, physiologists began 
in the nineteenth century to perform experiments upon isolated 
muscles and simple muscle activities of the intact organism, using an 
easily observable criterion of fatigue. Their purpose was to discover 

5 A. T. Poffenberger, Principles of Applied Psychology, New York, D. Apple- 
ton-Century Co., Inc., 1942, p. 111. 



[Ch. 3 





I .90 
I .80 






Intelligence Test 









1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 


Figure 2. Relation Between Output and Feelings in Four Forms of Mental Work 
Solid line represents output, and dotted line represents feeling record. 
(Reprinted from Poffenberger, op. cit., p. 111.) 

the fundamental processes and principles of muscular fatigue, in the 
belief that these discoveries would lead to better understanding of 
fatigue in more complex activities. The criterion of fatigue was a re- 
duced capacity of the muscle to perform, as a result of previous work. 


By thinking of fatigue in these objective terms, they were able 
to obtain definite advances in our knowledge of muscle function and 
of certain aspects of muscular fatigue. We shall presently review 
the results of these experiments upon muscle capacity. The same con- 
cept of fatigue has been applied to many activities outside of the lab- 
oratory, activities which are more complex than those which were the 
subject of these physiological investigations. 

As a yardstick for measuring fatigue of all kinds of work, and 
under all kinds of conditions, reduced capacity for work is subject to 
a number of difficulties. It is useful and justifiable to study fatigue 
in this manner when we are dealing with an isolated muscle in the 
laboratory, or even with carefully controlled muscular activities of 
an intact organism. When the same definition is applied to such 
complex activities as daily work in a factory or in the schoolroom, 
however, it is much more difficult to justify. The output may be 
readily observed, but it is often impossible to ascertain what has hap- 
pened to the level of interest of the worker. In short, we do not 
know what has happened to the capacity of the worker, only that his 
output has changed. Muscio ® has considered this difficulty so serious 
that he has proposed dropping the term ''fatigue" altogether. 

Another proposed solution of this problem is that fatigue be de- 
fined as reduced efficiency rather than as reduced capacity."^ In other 
words, if the worker has to exert greater effort to accomplish a given 
amount of work, he is becoming fatigued. If this definition is 
adopted, fatigue itself would await the solution of the general prob- 
lem of measuring input. Perhaps this is the only way in which these 
difficult problems of logic can be fully settled. At the present stage 
of our knowledge, however, fatigue itself refers to a group of ele- 
ments of cost which have not been analyzed and separated. Until 
that analysis is complete, it is preferable to use the term ''fatigue" as 
a general term for these elements, and to try to find some indicators 
of an indirect sort which can be used as additional measures of effort 
and added to the input factor in our efficiency ratio. 

Under laboratory conditions, as we have said, a definition of fa- 
tigue in terms of reduced capacity to perform is often workable be- 
cause capacity changes are reflected in output. With interested, 
cooperative subjects it is possible to use tasks which are extremely 
severe, and which require the subject to work nearly at capacity. It 

6 B. Muscio, Is a fatigue test possible?, British Journal of Psychology, 1921. 
12, 31-46. 

7 M. E. Bitterman, Fatigue defined as reduced efficiency, American Journal of 
Psychology, 1944, 57, 565-572. 



[Ch. 3 

is also possible to keep a rather close check upon the individual sub- 
jects and their behavior. They can be questioned and expected to 
answer quite frankly, in so far as they are able to describe their 
interest and degree of effort. 

Under field conditions in the factory or school, however, these 
conditions cannot often be met, and consequently measurement of 
fatigue in terms of output-decrement should be accepted only criti- 
cally and with reservations. Unfortunately, there have been some 
who have not recognized this need in evaluating studies of industrial 

Most of the preceding discussion was concerned with fatigue in 
muscular work. The precautions which are necessary in applying 
the concept of fatigue as reduced capacity to perform are even more 
important in the kinds of work which we speak of as "mental" tasks. 
Many muscular tasks make repeated demands upon a single set of 
muscles and associated nerves. It is under these conditions that we 
observe the most rapid and clear-cut decrease in performance. Most 
"mental" work is, however, extremely variable in the demands which 
it makes upon the body. Since these demands are less sharply lo- 
calized, the pattern of performance in these tasks does not show the 
regular decreases which are clearly seen in more limited muscular 
work. In mental work, it is only when we consider an extremely 
artificial task in which one small operation is repeated at frequent 
intervals that rapid and clear-cut decrement appears. It is likely that 
the dissimilarities between these artificial mental tasks and normal 
working activities are greater than the differences between work in 
the ergographic experiment and other types of muscular activity. 

To sum up this phase of our complex problem : Measuring fatigue 
by measuring the reduction in performance during work is subject 
to severe criticisms, but it is probably justified if used with care in the 
following situations : 

(a) Carefully controlled laboratory situations with highly motivated 

(b) Factory work or sports in which there is a heavy demand upon 
the muscular resources of the body. 

Laboratory Studies of Muscular Fatigue 

Techniques of Study. — When a muscle and its attached nerve are 
removed from the body of a cold-blooded animal like the frog, elec- 
trical shocks applied to the nerve or the muscle cause the muscle to 

Ch. 3] 



contract. It is possible to arrange apparatus which will make a 
graphic record of the muscle twitches, and to study the effect of a 
variety of factors upon the strength and duration of the contraction. 
These experiments have now become a common part of the laboratory 
exercises in college classes in physiology. It is found that the muscle 
which is stimulated repeatedly and frequently gradually diminishes 
its response until finally it will contract no longer. Fatigue is defined 
as the loss of strength of the muscle after these repeated stimulations. 

Figure 3. Finger Ergograph as Designed by Mosso 

The pointer at upper left traced a record of movements upon a smoked drum (not shown). 
The first and third fingers are held stationary by being inserted into metal cylinders. Lifting 
is done by the second finger. 

If the muscle is no longer able to respond at all, it is spoken of as 

Some of the results of these experiments upon isolated muscles 
have played a part in developing theories of fatigue which are current 
among physiologists. We shall therefore have occasion to refer to 
these results a little later. 

Similar experiments upon muscles in the intact organism were 
made possible by the instrument known as the ergograph.^ In this 
instrument (shown in Figure 3) ^ a weight is attached by a cord to a 

8 A. Mosso, Les lois de la fatigue etudiees dans les muscles de rhomme, Archives 
italiennes de Biologie, 1890, 13, 123-186. 
8 Ibid., p. 130. 



[Ch. 3 

fingertip, the hand, or the foot. The subject is instructed to raise 
the weight each time a signal is given, and to continue to raise the 
weight as high as he can each time the signal is heard. In this way a 
pattern of muscle contractions can be traced upon the recording drum 
in a fashion similar to that described in the experiments with isolated 
muscle. In this case, however, the method of stimulating the muscle 
is the normal 'Voluntary" nerve impulse rather than the artificial 

Figure 4. Records of Two Ergograms 

Each stroke represents one lift and drop of the weight. The temporal order is from right 
to left. The two records were made by different subjects to illustrate the characteristic 
individual pattern found by Mosso. 

electric shock. There is also another difference. In the intact body, 
a weight is not always lifted by exactly the same combination of 
muscles. There is, then, a possibility that the muscles involved in 
writing the ergographic record will not be the same throughout the 
experimmt. This is partially controlled by strapping down the limb 
of the subject and, if the finger is to be used in lifting, restricting the 
movements of the other fingers. 

The advantage of this approach to fatigue is obvious. There is no 
longer any possibility of differing meanings of the term, and fatigue 
becomes measurable. These advantages have been considered so 
important that a great deal of research upon more complicated activi- 

Ch. 3] 



ties, including factory work and school work, has employed the same 
definition in terms of decreased output. 

It is important to note, however, that there are certain definite 
conditions under which this approach to fatigue can be made. In the 
case of the isolated muscle, a reduction of the force of contraction, as 
shown on the record, means that the system no longer has the capacity 
to respond to the stimulus. Is this also true in the ergograph? 
Suppose, for example, that the subject simply loses interest in the 
task, and reduces his output for that reason. Should this also be 
called fatigue ? It does not seem useful to do so. Therefore we 
must specify that fatigue can he defined as a loss of strength of con- 
traction only under constant motivation. In a great many of the 
ergographic experiments, this condition has been reasonably well 
satisfied. The subjects have been men who were interested in the 
problem and definitely motivated to perform with their maximum 
possible strength. 

Evaluating the Results of Ergographic Experiments. — During 
the years around 1900 there was a widespread interest in research in 
fatigue by using the ergographic technique. Many of the results 
were helpful in increasing our understanding of the problem. Un- 
fortunately, the data which are reported from the experiments are 
extremely difficult to evaluate because the experimenters of that day 
paid little attention to the possible effects of chance in determining 
their results, and often failed to control sources of constant errors. 
In recent years there has been some loss of interest in this type of 
research, and many of the studies which have been carried out still 
omit statistical analyses. This is unfortunate, since our modern 
methods of control in experimentation, and of statistical analysis, 
could clarify many things which were left obscure by these experi- 
menters. We have, as a result, many suggestive findings, but little 
that is conclusive. In spite of this, many of these results have been 
adopted in an uncritical way by texts in physiology and applied psy- 
chology. We shall present some of these findings because of their 
interest, but we do so only after a preliminary warning. 

Here is a typical case which illustrates this warning. The Belgian 
Mile. loteyko examined the effect of exhaustion of the muscles of a 
finger of one hand upon the muscular strength of the opposite hand.^** 
This experiment had a number of theoretical consequences relative to 

10 J. loteyko, L'effort nerveux et la fatigue, Archives de Biologic, 1899-1900, 
16, 479-535. Participation des centres nerveux dans les phenomenes de la fatigue 
musculaire, L'Annee de Psychologie, 1900, 7, 161-186. 



[Ch. 3 

the nature of fatigue. Out of the eighteen subjects, some showed a 
decrease in strength of the opposite hand, others showed an increase, 
and the remainder showed no change. Our first reaction should be 
to raise the question whether this result represents a simple matter 
of chance variations in strength of the hand, variation which would 
have occurred regardless of the fatigue in the right hand. loteyko, 
however, does not consider this question at all. Instead, she says that 
there are three types of individual. In one type, the fatigue is mostly 
local and in the muscles themselves, and there is no transfer of effect 
to the opposite side. Another type has fatigue which affects the 
central nervous system more severely, and therefore diminishes the 
strength of any reaction. The third type is ''mixed." We are not 
concerned at the moment with these theories. There are many criti- 
cisms which could be raised against them even if we accepted 
loteyko's first assumption. The first assumption itself is not, how- 
ever, demonstrated. In order to demonstrate it, it would be necessary 
to show that certain individuals consistently increase the strength of 
one hand after fatiguing work with the opposite hand, while others 
consistently decrease their strength. In other words, we need to 
show that we have something more than a random variation in 
strength unrelated to the effects of fatigue in the opposite side. This 
cannot be done without adequate statistical and experimental controls. 
The experimenters of that time were not aware of the methods which 
are now in routine use for answering questions of this kind.^^ 

In addition to these problems of experimental method, there are 
certain assumptions sometimes involved in the use of the ergograph 
which must be clearly realized in interpreting the results. The most 
important is that "exhaustion" can be taken as a fixed point of ref- 
erence — as representing a fixed state of the muscles, or a given 
"quantity" of fatigue. 

Let us see how this assumption is important. Suppose that we are 
trying to find out what load or weight leads to most efficient per- 
formance of the finger muscles. The method would be as follows : 
Let the subject work to "exhaustion" a number of different times 
with each of several weights. We then find the amount of work done 

11 More recently, Collier has found a statistically significant crossing of fatigue 
in a task involving rotating the forearm at maximum speed. (R. M. Collier, The 
crossed effects upon voluntary movement of a unilaterally induced fatigue, Journal 
of Experimental Psychology, 1938, 23, 26-43.) The number of subjects and the 
methods of analysis were not adapted, however, to determining whether there were 
different individual "types" of subject. The crossed fatigue resulted in decreased 
speed, but also in increased amplitude of movement. The greatest effect involved 
the speed of reversing the direction of movement rather than the speed of move- 
ment itself. 


in each ergogram by multiplying the weight by the total distance it 
has been lifted. It will be found that the total amount of work varies 
from weight to weight, and in this manner we determine the optimal 
point. As we should expect, a weight which is too heavy exhausts 
the muscles so quickly that very few lifts will be made. On the other 
hand, with a very light weight, little is accomplished on each lift, and 
the slower fatigue is canceled by this fact. 

To simplify the underlying procedure : we have assumed that the 
cost of work per ergogram is a constant, that a given amount of fa- 
tigue is always involved. If this is true, the conditions which produce 
the greatest total accomplishment in an ergogram are the most effi- 
cient conditions. This is often taken as a self-evident assumption — 
an axiom, but it really depends upon the question of whether 
exhaustion or cessation of movement always implies the same state of 
the organism. This must be discovered before the procedure indi- 
cated above is justified. 

There is, however, evidence that the effects of completing an 
"ergogram" upon the organism are not always the same, and that we 
are not justified in saying that the cost remains constant from one 
ergogram to another. Of course, this evidence itself is incompletely 
established for reasons already mentioned, but it is enough to show 
that the assumption is not acceptable as an axiom which needs no 
testing. Here is one sort of fact which argues against the assump- 
tion: Fere,^^ one of the early workers, says that fatigue is more 
"profound" under some conditions than under others. For example, 
he finds that the total work done before exhaustion, increases as the 
rate of contraction is decreased. But he also finds that the recovery 
is more rapid when the exhaustion was reached by a series of rapid 
movements than it was when it resulted from a series of slower 

In order to decide upon the relative efficiency of various conditions 
and methods of muscular work, it is therefore necessary to take into 
consideration a number of factors : ( 1 ) Was the motivation constant, 
so that exhaustion represents the same stage in ability to contract 
under each condition? (2) What effect does the condition have upon 
recovery, as well as upon the amount of work done up to the point of 
"exhaustion?" (3) What is the effect of the condition upon a long 
series of ergograms separated by rest periods, as well as upon a single 

12 Charles Fere, Travail et plaisir, Paris, 1904. 

13 Some of the more recent experimenters have been able to produce very inter- 
esting hypotheses about the nature of muscular fatigue by analyzing variations in 
recovery as well as output of the ergogram itself. 



[Ch. 3 

ergogram or a few? The latter question concerns any cumulative 
effects of the work which do not show up in short term efforts; it 
may be extremely important. 

Fatigue and Mechanical Efficiency. It might appear that 
measurement of fatigue and measurement of metabolic cost are 
simply two different approaches to the same effects of work, espe- 
cially if we are concerned with work which is largely muscular in 
nature. At one time Atzler made this assumption explicitly, stating 
that a method of work which permitted the greatest output before 
exhaustion would also be the most efficient from the point of view 
of metabolic cost. He cited experiments in which he had his subjects 
work at various rates upon the bicycle ergometer. Those speeds 
which permitted the greatest amount of work to be accomplished 
before the subject was exhausted were also the speeds at which me- 
chanical efficiency was at a maximum during the early minutes of 
work. That is, the lower the oxygen consumption at the beginning 
of the work, the more the subject accomplished before he was ex- 

Simonson, however, questioned the possibility of generalizing 
this result into an axiom.^^ In fact, he was able to show that factors 
other than variation in speed could produce an opposite effect. In 
lifting weights with the outstretched arm, the weights which were 
handled most efficiently from' the point of view of oxygen consumj>- 
tion were the ones which were the least efficient from the point of 
view of total accomplishment in the ergogram. The results corrob- 
orated those of Atzler when the changes in efficiency were produced 
by changes in speed of work, but contradicted them when the load 
was varied. 

If metabolic cost and fatigue expense are not correlated in heavy 
muscular work, as Simonson's experiments indicate, they must be 
measures of independent elements of input, and so must be measured 
separately. If this is true for muscular work (where there was some 
reasonable expectation that they would be identical), it must be even 
more evident when we consider tasks involving much less heavy 
muscular effort. 

Summary of Research on Muscular Fatigue Because of the 

difficulties of research methods discussed in the preceding paragraphs. 

1* E. Atzler, Die Bekampfung der Ermudung, in E. Ludwig's Der Mensch im 
Fabrikbetrieb, Berlin, 1930, 15-37. 

15 E. Simonson, Der heutige Stand der Theorie der Ermudung, Ergebnisse der 
Physiologie, 1935, 37, 303-305. 

Ch. 3] 



views and theories of the nature and basis of muscular fatigue vary 
considerably. The following highly condensed analysis of the im- 
plications of these researches must, therefore, be tentative in char- 
acter, although we have tried to use the more fully verified results. 
We shall attempt an organized summary because the only other al- 
ternative is a listing of a great number of heterogeneous results and 
theories which would have little meaning. Even if the following 
statements are tentative in character, it is hoped that they will give 
the reader some understanding of the scope of the problems facing 
specialists in muscular fatigue, and some notion of the progress which 
has been made to date. We must condense our summary to a very 
small compass in relation to the number of experiments and discus- 
sions available, because a more complete survey would require space 
and time out of proportion to the practical gains. 

1. Loci of Muscular Fatigue. A question which has con- 
cerned most of the researchers in this field is: ''When capacity to 
continue performing a given task has been reduced or eliminated 
altogether, which parts of the total mechanism are primarily respon- 
sible for the loss — the muscles themselves, the peripheral nerve con- 
nected with the muscles, or special portions of the central nervous 
system?" Since all these mechanisms are engaged in the perform- 
ance, they are all affected. It appears, however, that the effects are 
more serious in some places than in others. The relative importance 
of the events in various parts of the total mechanism varies, how- 
ever, with the conditions and type of work. 

(a) Local ^'Dynamic'' Work. We may consider first the "local** 
fatigue which occurs when a small group of muscles is involved, as 
in the finger-ergograph. It appears that the central nervous system 
is responsible for reduction of output before the peripheral mecha- 
nisms are seriously affected in most kinds of localized muscular effort 
of the intact organism. The muscle itself is injured only under ab- 
normal experimental conditions, or in those experiments where the 
muscle is isolated from the rest of the body. This fact is of impor- 
tance not only because of its contribution to our understanding of 
the total pattern involved in fatigue, but also because chemical 
theories of the basis of fatigue have in the past been based upon the 
isolated muscle, and are therefore concerned with an abnormal or 
atypical kind of fatigue, or one phase of fatigue which is of less 
importance than others. 

After exhaustion of the power to lift a weight by voluntary con- 
traction, the muscle will still respond if electrical shocks are applied 



[Ch. 3 

to the muscle directly or to its nerve/® It has been found, however, 
that after voluntary exhaustion in work with a heavy weight and a 
very fast pace, there is a definite effect upon the power of the muscle 
to contract under electrical stimulation, while exhaustion produced 
by lighter loads or slower contractions (slower fatigue) shows no 
evidence of fatigue in the muscle itself/^ In fact, Reid found that 
he could not produce fatigue by direct stimulation of the muscle or its 
nerve unless the stimulation was repeated over a long period of time, 
or unless he used continuous stimulation rather than periodic shocks. 
This applies, of course, to the muscle in an intact organism. The 
rapid fatigue under this kind of stimulation in an isolated muscle, 
must, he believes, be due to the lack of normal blood circulation. 

(b) Postural Restriction — Maintained Contraction. In static 
work (holding a weight continuously), as contrasted to dynamic 
work (repeated contractions of short duration like those referred to 
in the preceding discussion), the roles of central and peripheral fa- 
tigue may be different. In his comparisons of voluntary contractions 
with direct electrical stimulations, Reid concluded that most of the 
reduced capacity is a result of central factors in both static and dy- 
namic work. Simonson, however, advanced the theory that static 
work with very heavy weights which produce rapid exhaustion is 
largely a matter of local muscular effects, while slowly developing 
static fatigue involves a more marked central component. His evi- 
dence was largely indirect and has been questioned by more recent 
researchers, however. Kurbatova and Sheidin reach an opposite con- 
clusion on the basis of the effects of rest.^® They find that work 
capacity recovers much more rapidly after quick exhaustion in static 
work than it does in the slower development of fatigue. This re- 
covery from rapid static fatigue takes place even if the circulation of 
the arm is cut off during the rest period. Massaging the muscles is 
more effective following slow fatigue than it is after rapid fatigue. 
More recovery from the rapid fatigue could be produced by some 
activity of the other arm than by complete rest. These conclusions 
would therefore be exactly the reverse of these drawn by Reid for 
dynamic work. In static work, the central factors are more im- 
portant in rapid fatigue than they are in slow fatigue; in dynamic 
work the central factors are most significant in slow dynamic work. 

16 Mosso, op. cit. 

i^C. Reid, The mechanism of voluntary muscular fatigue, Quarterly Journal of 
Experimental Physiology, 1928, 19, 17-42. 

18 Simonson, op. cit., p. 330. 

19 I. N. Kurbatova and Y. A. Sheidin, On the question of fatigue in static work, 
Uchenie zapiskie, Leningrad University Publications, 1938, Vol. 6, No. 23, 149-163. 



Little can be said at this time about the nature of the changes in 
the central nervous system that are responsible for diminishing mus- 
cular output; the role of the central nervous system in fatigue v^^as 
discovered by a process of elimination. The portions of the nervous 
system involved might be sensory centers, motor centers, or co- 
ordinating connections. There are, of course, various theories, but 
the evidence is fragmentary, and this is a problem which needs much 
further analysis. It is known that fatigue of the nerve fibers them- 
selves is a negligible factor under normal conditions. Solution of the 
problem of the intimate nature of the changes involved awaits better 
understanding of the physiology of the nervous system in general. 

At any rate, the multiplicity of factors involved even in what ap- 
pears to be a very simple kind of "local" muscular fatigue, and the 
variability of the factors with changes in conditions, should make it 
clear that practical recommendations upon the control of fatigue 
cannot be worked out by derivation from any general theory of the 
nature of fatigue. Even if the general theory were well established, 
we should still need to know what combination of factors is important 
in any particular case before fatigue could be controlled. 

In more generalized muscular work of a heavy kind (dromal 
tasks), such as running, bicycling, carrying heavy loads, and the like, 
the nature of the basis of fatigue is probably still more complex. It 
has been concluded that in rapid exhaustion in heavy work, the pri- 
mary factor is a failure of the circulatory and respiratory mecha- 
nisms. The central nervous system may be affected by the deficiency 
of oxygen, however, and consequently may again be a central -factor 
in the cessation of effort. In general muscular exertions which are 
of the slower kind — muscular drudgery in which a steady effort is 
continued for longer periods of time, as in hiking, digging ditches, 
other common labor, and the like — the mechanisms are probably 
different from those involved either in ''local" fatigue or in rapid 
exhaustion in heavy work. It is, in fact, probable that the mecha- 
nisms are so distinct that the basis of fatigue could not even be sug- 
gested from experiments upon the other kinds of work. 

2. Rest and Recovery. Some of the factors which affect the 
rate of recovery have already been mentioned in discussing the loci 
of fatigue. Other facts, of interest in their own right, will be sum- 
marized briefly at this point. Amount of recovery is usually meas- 
ured in terms of the total amount of work done when the work is 
taken up again following a period of rest. That is, following a given 

20 Simonson, op. cit., p. 340. 



[Ch. 3 

period and condition of rest, the task is resumed, and. the subject 
works until he reaches complete exhaustion. In the case of dynamic 
work, the amount of work done during this second period of work is 
obtained by multiplying the weight by the total distance through 
which it is lifted — the sum of the heights of all the contractions. 

Using this method, it was found that the later portions of a period 
of work are much more costly than the initial period. Maggiora, for 
example, measured the time required for complete recovery of the 
activity, i.e., the time at which the second ergogram showed the same 
total output as the first.^^ In another series of experiments, he al- 
lowed the subjects to work only half as long as was normally neces- 
sary to reach exhaustion, allowed them to rest, and then determined 
the length of rest necessary to produce a normal ergogram. Under 
these latter conditions the time required for complete recovery was 
only one-fourth that required for complete recovery following ex- 

The recovery following exhaustion does not follow a smooth 
trend, but appears to be cyclical. A short rest (ten to fifteen min- 
utes) may sometimes produce more recovery, as measured by output, 
than a longer period.^^ Whether or not this stimulation of perform- 
ance following a brief rest is to be considered true recovery, or some 
kind of spurious effect, depends upon our definition of "recovery." 
The definition used in these experiments was stated above. In terms 
of this definition, the subject does recover, and then goes into a slump 
again before attaining permanent return to the normal condition. It 
is still possible, however, that the mechanism involved in this first 
rapid recovery is different from that which produces the final and 
slower return to normal capacity. It is also possible that perform- 
ance is more costly after the brief rest period, even though this cost 
is not reflected in performance at the time. 

Recently, Sharp has thrown some interesting new light upon the 
rapid and temporary recovery in the first few minutes after ergo- 
graphic work.^^ By analysis of muscle potentials from the active 
muscles during the rest period, he found a spontaneous recovery of 
muscle tension at about the time of the first temporary recovery in 
work capacity. The pattern of potentials is shown in Figure 5, taken 
from Sharp. It would appear that muscle tension, or tonus, is cor- 

21 A. Maggiora, Les lois de la fatigue etudiees dans les muscles de rhomme: 
II, Archives italiennes de Biologic, 1890. 13, 187-241. 

2 2 Charles Fere, Travail et plaisir, Paris, 1904. C. W. Manzer, An experi- 
mental investigation of rest pauses. Archives of Psychology, 1927, No. 90, 1-84, 

2 3 L. H. Sharp, Effects of residual tension on output and energy expenditure in 
muscular work, Journal of Experimental Psychology, 1941, 29, 1-22. 

Ch. 3] 



related with the capacity of the muscles to perform. What this ten- 
sion means with respect to the state of the mechanism of action, and 
what brings about the change in tensions, still remain to be dis- 
covered, but Sharp's findings represent one more clue to the nature 
of the process of recovery. 






vl 400 











M/^R-SK/N ReSlS- 




95 5 

/O /5 SO Z5 


Figure 5. Skin Resistance and Muscle Potentials Following Ergographic Work 
(From Sharp, op. cU., p. 7.) 

Reid believes that the process of recovery is influenced by proprio- 
ceptive (muscle sense) stimuli in the muscles, even when the principal 
locus of fatigue is in the central nervous system.^* For example, he 
finds that electrical stimulation of the muscle during the "rest" period 
— i,e., when the muscle is not being used in voluntary contractions — 
slows down recovery. His interpretation is that the sensory impulses 
maintain the central fatigue condition. In other words, the stimula- 
tion of the muscle was such that it would not produce appreciable 
fatigue of the muscle itself, and therefore should not affect recovery 
on that basis. It may well be that Sharp's findings on tension, and 
Reid's conclusions, are both related to a similar factor in recovery. 

2* Reid, op. cit. 



3. Effect of Load and Rate of Performance. For a con- 
stant rate of work — a fixed number of lifts of the weight per minute 
— there is an optimal load. Weights heavier or lighter than this op- 
timal load will not permit the subject to accomplish as much before he 
reaches exhaustion. With the lighter weights, fatigue progresses 
more slowly, but not so much work (as measured in foot-pounds) is 
accomplished in each lift of the weight. For loads which are too 
heavy, fatigue is so much more rapid that it more than overcomes 
the advantage due to the greater accomplishment in each lift of the 
weight. In general, the slower the rate, the heavier the load for the 
optimum combination. 

With a fixed load, the total accomplishment before exhaustion, in- 
creases as the rate is slowed down. It is possible to leave sufficient 
time between lifts so that the work can be continued almost indef- 
initely.^^ Ultimately, however, there will come a point where the 
subject is incapable of continuing. Performance may remain con- 
stant for a long time and then suddenly drop to zero.^^ 

4. Fatigue and Spread of Activity. As fatigue increases, 
there is a spread of activity to other muscles of the hand. Ash found, 
for example, that it was possible to record contractions in the fingers 
which were not actively lifting the weight. As the ergogram pro- 
gressed, contractions of these muscles became greater and greater.^' 
If the middle finger is "exhausted" while the other fingers are strapped 
in place, the weight can again be lifted if these unused fingers are 
released. Ash concludes from this that fatigue involves a loss of fine 
control over these muscle co-ordinations. 

5. Theories of the Chemical Basis of Fatigue. While most 
physiological texts and general summaries of the subject of muscular 
fatigue customarily devote a great deal of space to the chemistry of 
muscle contraction, the importance of this topic for our purposes is 
quite small. We shall seek only to place this topic in relation to the 
other phases of fatigue already discussed. 

A number of facts derived from the study of fatigue in the isolated 
muscle, and some of the findings related to fatigue in the intact or- 
ganism, led to the development of theories of fatigue in terms 
of ''fatigue substances." These were end-products of the process of 
muscular contraction which were held responsible for the loss of 

2 5 Maggiora, op. cit. 
2 6 Fere, op. cit. 

27 I. E. Ash, Fatigue and its effect upon control, Archives of Psychology, 
No. 31, 1914. 

Ch. 3] 



capacity of the muscle to perform. Lactic acid is one of the sub- 
stances which has received considerable attention from this point of 
view.^^ Increased acidity was supposed to aifect the nerve junction 
with the muscle so that it no longer transmitted impulses ; in higher 
concentrations, to affect the power of the muscle itself. 

There is no doubt that an understanding of the chemistry of mus- 
cular contraction has its place in completing the total picture of the 
economy of activity. In heavy muscular work, the extra oxygen 
taken into the body is employed in oxidizing these chemical products 
of activity and in restoring the muscle to its normal chemical state. 
In the isolated muscle, and perhaps in extreme conditions of the intact 
organism, these activity products are probably in sufficient concen- 
tration to decrease the capacity of the muscular apparatus to perform. 
Even under more normal conditions, products such as lactic acid 
spread to other parts of the body through the blood stream, and may 
have some indirect effects in this manner. 

The production of lactic acid is now considered, however, only one 
among a great many different bodily events involved in muscular fa- 
tigue. Even in isolated muscles it is probably not the sole factor 
leading to a reduction in capacity. As we have seen, the emphasis 
has shifted from peripheral muscular events to the central mechan- 
isms of control in the nervous system. 

Summary of Muscular Fatigue — There is no doubt that research 
into muscular fatigue has made progress. A part of the picture has 
begun to emerge, although its details are still hazy. It is a gain to 
know that muscular fatigue, even under the simplest conditions, can- 
not be subsumed under some simple theory or formula. The defini- 
tion of fatigue in terms of reduced capacity to perform, while it is 
difficult to apply in a factory or under other practical circumstances, 
is workable and useful for physiological investigations into the basis 
of fatigue. 

In the light of the results we have mentioned, it cannot be as- 
sumed that fatigue in mental work will be understood by merely 
extending the conceptions derived from muscular work. Even in 
muscular work there are probably many different conditions which 
would be classed as "fatigue states" in that they reduce capacity for 
performance, and in that these states result from previous exercise. 
How much these states have in common beyond these gross features 
still remains to be determined. Similarly, it is likely that still dif- 

2 8 See Atzler, op. cit. in Ergebnisse der Physiologic, 1938, 40, 1939, 41, and 
Simonson, op cit., for general reviews of this topic. 



[Ch. 3 

ferent states underlie fatigue in mental work, and we know still less, 
if that is possible, about what they have in common with the con- 
ditions underlying muscular fatigue. 

It seems likely that there are several important bodily changes 
which occur in a variety of combinations in various fatigue condi- 
tions. We cannot assume in advance, however, that fatigue in mental 
work is nothing more than a new combination of the same factors 
involved in muscular fatigue. It is much safer to approach problems 
of mental fatigue directly, without preconceptions. If some of the 
factors involved in mental fatigue are similar to those in muscular 
fatigue, the fact should appear in the results of research. If there 
is little similarity, the direct approach to mental fatigue will have 
advanced us further, since it would prevent our embarking upon a 
dead-end trail. 

As for measuring fatigue as an element in the cost of muscular 
work, decrement of performance in highly motivated workers is still 
the only useful indicator. This, as we have pointed out, has its draw- 
backs when applied to the practical setting because of variation in 
motivation. Nevertheless, research into the physiological basis of 
fatigue has not provided us as yet with any adequate bodily test for 
determining the degree of fatigue that results from a given task. 
Such a test may ultimately be perfected, but meanwhile we must make 
the best of output decrement in the study of muscular work. 

Chapter 4 


Kinds of Sedentary Work. — As mentioned previously, it is con- 
ventional to divide all work into the two main classes of "mental" 
and ''muscular." We have found that it is still useful to adopt a two- 
fold classification for purposes of chapter division, even though the 
traditional division of the two classes is not very satisfactory. In the 
preceding chapter we adopted a revised conception of the class of 
muscular work, including only those classes of work in which the 
central feature of the task is the development of muscular force. In 
other words, it is not merely that muscular activity is involved in the 
work, but that the stress is upon the human body as a machine for 
performing work in the physical sense (''dynamic" work) or to main- 
tain equilibrium of a physical system ("static" work). 

We have indicated that the measures of cost of work which are 
useful for studies of muscular work are of doubtful value when ap- 
plied to other tasks commonly described as "mental." Before taking 
up the measures of cost of work for this latter kind of task, it will be 
well to consider the kinds of activity that are to be studied. So far 
we have characterized sedentary work primarily in a negative way — 
the class of work which is left after muscular work has been ab- 
stracted. No term is entirely adequate to describe this class, but the 
term sedentary seems preferable to mental because many of the tasks 
we shall include are not commonly thought of as "mental," and the 
implication that any activities take place "in the mind" is certainly to 
be avoided. All activities are bodily activities. 

We offer here a tentative classification of types of sedentary work. 
The classification is given here mainly to indicate the kind of subdi- 
vision which may be necessary in order to bring greater meaning into 
our facts. This particular classification may not be the most useful 
one for other purposes. It is especially likely that the headings are 
too broad and that they need further refinement. Nevertheless we 
believe that it is helpful to bear such a pattern in mind as we examine 
the results of researches which investigate the cost of work in the 
general area of sedentary or mental work. 




[Ch. 4 


Kinds of Sedentary (Mental) Work* 

1. Problem Solving (With minimum muscular and sensory involvement.) 

e.g., calculation, solution of mathematical problems, composing, plan- 
ning, directing the work of others, supervising, socialized tasks 
such as selling. 

2. Continued Sensory Adjustment (Primarily visual tasks.) 

e.g., proofreading, visual inspecting, radio code reception, piano tun- 
ing, and the like. Reading — under difficult conditions or when 
long continued. 

3. Motor Skill (With patterning and accuracy of movement as the central 
core, and force only an accessory feature.) 

e.g., typing, drawing, various machine operations, assembling machines, 
woodworking, sewing, acting, speech-making. 

4. Sedentary-muscular (Light muscular tasks with little skill or control 

e.g., watchman, crossing guard, machine feeder, or any task where the 
main requirement is that the proper movements be made at the 
proper time. The movements themselves have few elements of 
skill or of force involved. 

* Modified from T. A. Ryan, Varieties of fatigue, American Journal of Psychology, 1944, 
57,^ 565-569. 

The distinctions between these classes of work should be evident 
from the examples given. Many tasks may consist of alternations 
between the different types (including muscular), or of various com- 
binations of the above features. For example, in long-continued 
typing, postural factors and even localized muscular strain of the 
fingers may enter. These features alone, however, would be con- 
sidered only incidental elements of the task in most cases, and the 
task would be classified according to the main features. 

We shall see that laboratory studies of fatigue and effort have 
stressed Types 1 and 2 — problem solving and continued sensory ad- 
justment. Occasionally a simple kind of motor skill has been investi- 
gated, not as a separate problem but grouped with Type 1 as an 
instance of ''mental" work. Thus tasks involving motor skill have 
not received the attention which is consonant with their importance in 
practical jobs, except for gross statistical analyses under factory con- 
ditions. The situation for this kind of work is therefore very similar 
to that for the "steady grind" mentioned in the chapter on muscular 

Investigations of continued sensory adjustment have been re- 
stricted to the problem of visual fatigue. Here the investigation has 

Ch. 4] 



been along* special lines, and in consequence there is little confusion 
with other kinds of work. In fact, the problems and methods are so 
specialized that we shall not discuss them in the present chapter, re- 
serving this topic for a later analysis of the problems of lighting 
(Chapter 5). 

When all the facts are in, it may be found necessary to regard 
several of these classes as distinct and independent. Work involving 
muscular skill, for example, may be found to affect the organism in a 
manner quite distinct from problem solving, even though both lay 
principal stress upon control, direction, and timing of activity. Per- 
haps too, the steady grind mentioned as a kind of muscular work 
belongs with these skilled tasks rather than with the other forms of 
muscular work. Another possibility is that some tasks involving 
muscular skill resemble problem solving, others the steady grind. 
The various kinds of work called sedentary are therefore grouped 
together only tentatively, pending further information and following 
the groupings inherent in the thinking of past researchers. 

In spite of these doubts of the basic scientific validity of the two- 
category scheme, the class of sedentary tasks does possess two at- 
tributes which are worthy of note. First, all are tasks in which the 
energy cost of work is an inadequate indicator of the total cost of 
work to the worker. Second, the class includes the vast majority of 
tasks which are important in the modern working world. As time 
goes on, muscular tasks assume smaller and smaller importance in 
industry, maintaining their position only in the world of sports. 

Fatigue and Decrement in Sedentary Tasks 

In the preceding chapter we discussed some of the difficulties in- 
herent in the use of output decrement as a measure of fatigue. These 
difficulties are, if anything, even more troublesome in studies of 
sedentary work. Nevertheless, decrement has been used in studies of 
fatigue in sedentary work of many kinds both in the laboratory and in 
the practical setting of the factory or school. Our first task will 
therefore be a more detailed analysis of the value of output decre- 
ment as an indicator of the cost of sedentary or mental activities. 
After surveying the results of laboratory studies of decrement, we 
shall look at some of the analyses of work curves in factory work. 

Factors in Decrement of Sedentary Work. — It was hoped that 
the investigation of the ergographic task under controlled conditions 
would lead to general principles of muscular fatigue which could 



[Ch. 4 

then be applied to more complex activities in daily life. Similarly, 
several experimenters have studied tasks of the kind roughly classed 
as "mental" in a hope of finding rules which could be applied to the 
office, schoolroom, or factory. 

The method of experimentation is essentially similar in the two 
cases. In the studies now to be considered, the subjects are asked to 
cancel letters in a page of text according to specified instructions, to 
add long columns of numbers, to react quickly to signals, and the 
like. With these tasks it is found that the rate of decrease in per- 
formance is much slower than it is with the ergograph. In fact, after 
several hours on some of these tasks, highly motivated subjects may 
have slowed down only slightly. For this reason it is much more 
difficult to find clear-cut changes which accompany the progress of 

Perhaps the greater difficulty in following the effects of prolonged 
mental work is due to the fact that there is a strong possibility of 
confusing several distinct phenomena. First, there are probably two 
or more different kinds of fatigue which change their relative im- 
portance with changes in working conditions. It is likely, for ex- 
ample, that a strong emotional strain going along with the work is 
one thing responsible for the feelings of tiredness of the worker. The 
bodily effects of this factor are probably not the same as the effects 
of the work itself, unaccompanied by their emotional factor. This 
factor is given more extended consideration in Chapter 9. Then, too, 
there is a muscular fatigue which enters in because of the restrictions 
upon movement which are required by the task. 

In addition to the three factors already mentioned, factors which 
are commonly grouped under the heading of fatigue, there is a fourth 
element which has been confused with them in certain experimental 
studies of work decrement in mental tasks. Yet it is considered by 
the worker himself as quite distinct from fatigue. This is the factor 
of monotony or boredom. We shall discuss monotony later in the 
general context of motivation in work, rather than in the current 

A fifth important factor affecting the pattern of output is practice 
or learning, which tends to counterbalance the effect of fatigue. In 
the early stages of an experiment, practice can have such a marked 
effect that performance improves continuously throughout the course 
of a day, in spite of any fatigue which may occur. One method of 
controlling this factor in studies of fatigue is to start the main part 
of the experiment only after the subject no longer shows evidence of 
improvement due to practice. There is often a long wait before this 

Ch. 4] 



Stage is reached, and even so, it is always difficult to be certain that 
the practice effects have been fully eliminated. 

Various tricks in the analysis of data have been designed in order 
to allow for the practice effect. One device is to compare the low 
point of performance in a given session with the maximum point of 
the succeeding period of work, following an extended rest.^ The as- 
sumption would be that the rest is long enough to eliminate the 
effects of fatigue, while the effects of practice carry over to the fol- 




>^ __ - 

1 1 1 




2 4 6 



2 4 6 8 




Figure 6. Hypothetical Effects of Fatigue and Practice 

When practice covers fatigue effects, fatigue may be measured by (M2-W2) instead of 
actual decrement (Mi-Wi). For simplicity, this drawing represents the case where there is 
no warming-up effect. 

lowing session. This method evidently does not allow for possible 
loss of skill or forgetting during the rest interval.^ Thus the decre- 
ment in performance, computed in this manner, would be less than 
the actual decrement in capacity occurring during a given session of 
work. (See Figure 6.) 

Other methods are still more complex and involve a great deal of 
experimentation devoted to the securing of control over factors like 
practice. Since most of the methods involve some assumption con- 

1 E. L. Thorndike, Mental fatigue, Journal of Educational Psychology, 1911, 2, 

2 T. Arai, Mental fatigue, Teachers College Contributions to Education, 1912, 
No. 54, p. 85. 



[Ch. 4 

cerning the nature of forgetting during a rest interval, it would 
appear much better to work with highly practiced subjects when- 
ever that can be done. Highly practiced subjects would exhibit 
greater uniformity of performance and more reliable results than 
beginners, as well as provide for more adequate control of practice 

There are also other factors which tend to counteract the influence 
of fatigue, upon decrement. Examples of these are the 'Svarming- 
up" at the beginning of work, and ''spurts," especially when the 
end of the work period is anticipated by the subject. These factors 
can be allowed for with relative ease, although they must be recog- 
nized and taken into account in any study of fatigue. 

Because of all the factors mentioned, there are few results among 
these studies of mental work which are clear enough to be of value 
in understanding the problem. Decrement in performance can be 
produced experimentally, although frequently the decrement is 
very small relative to the feelings of fatigue reported by the subjects.^ 
In a continuous alternation of tasks (tests of reaction, co-ordination, 
tapping, color naming, etc.) HoUingworth found some test perform- 
ances improved during a twelve-hour day, while others show^ed de- 
crease.* Even the decreases were relatively slight in amount (17 
per cent loss in opposites was the greatest decrement). Practice was 
controlled, but there was no continuous effort in any one task. Our 
earlier discussion of feelings of fatigue referred to Poffenberger's 
curves of performance and feelings for various tasks. Here some 
performances improved during the five and a half hours of the tests, 
an improvement attributed by Poffenberger to practice.^ 

These examples serve to show the great variability of changes in 
output with varying conditions and tasks even under laboratory con- 
ditions. Robinson ® has attempted to state the conditions which lead 
to decrement in a mental task, and in an experiment ^ which he carried 
out with Bills has demonstrated the importance of two of these 
factors in producing a rapid decline of performance. These two 
factors are homogeneity and conflict. 

3 See Arai, op. cit., for examples of decrement in long periods of arithmetical 

* H. L. HoUingworth, Variations in efficiency during the working day, Psycho- 
logical Review, 1914, 21, 473-491. 

5 A. T. Poffenberger, Principles of Applied Psychology, New York, D. Apple- 
ton-Century Co., Inc., 1942, pp. 110-111. 

6 E. S. Robinson, Principles of the work decrement. Psychological Review, 1926, 
33, 123-134. 

E. S. Robinson and A. G. Bills, Two factors in the work decrement. Journal 
of Experimental Psychology, 1926, 9, 415-443. 

Ch. 4] 



The first of these factors is evident in ergographic work as well. 
In fact it is commonly said that the reason that local muscular exer- 
cise shows rapid decrement, while many mental tasks show little loss, 
is due to the homogeneity of the ergographic task. Almost the same 
movement is carried out repeatedly with only brief time intervals 
between the reactions. Essentially the same peripheral organs are 
involved, and it may be that the central neural processes undergo 
little change from one reaction to the next. Robinson and Bills found 
that a ''mental" task which made similar demands would also produce 
a very rapid decrement. For example, the repeated writing of the 
same letter or pair of letters leads to a rapid reduction in output. 

The decrement in this co-ordinated task is different, however, 
from the decrement in ergographic work. Instead of a gradual 
diminution of strength of movement, performance is slowed down in 
writing as the result of frequent "blocks" when the subject is unable 
to respond at all for a brief time. Between blockings there seems to 
be little reduction in the speed of movement. The movements them- 
selves are rapid, but they are interrupted. The effect of homogeneity 
upon this co-ordinated skill is quite different, therefore, from that 
in ergographic work, and only in a very general way can it be said 
that homogeneity is a common factor in the two types of decrement. 
Another way in which the effect of repetition differs in the two kinds 
of task is that the letter-writing declines rapidly for a time, and then 
tends to level off. Other activities which involve speed as the measure 
of performance have also shown a similar pattern. It is clear that 
such tasks are not at all comparable to a task where performance is 
measured in terms of strength, while speed is controlled by the ex- 
perimenter. Therefore, while it can be said that homogeneity does 
lead to decrement in a mental task, it is impossible to draw a parallel 
from its effect upon muscular work. 

Bills has also cited other results of experiments upon mental work 
as illustrations of the role of the factor of homogeneity.^ He men- 
tions Poffenberger's finding that several hours of adding pairs of 
digits led to a 20 per cent loss by the end of the period, while an- 
swering questions on intelligence tests for the same time showed an 
improvement of 20 per cent. The greater variability of activity in 
the intelligence test is presumed to account for this difference. 

Homogeneity, then, is quite obviously one of the characteristics of 
a task which leads to rapid decrement. It is possible, however, as our 
comparison with ergographic work indicated, that the effect of ho- 

A. G. Bills, The Psychology of Efficiency, New York, Harper & Bros., 1943. 



[Ch. 4 

mogeneity may vary with the conditions and with the tasks involved. 
Although it was apparently not a problem in the experiments we have 
been discussing, homogeneity may also have an indirect effect by pro- 
ducing monotony and reduced motivation, as well as the more direct 
effect of reducing the capacity of the organism to go on with the 
task. Bills, for example, discusses the effects of homogeneity upon 
blocking in terms of a "refractory phase" of the mechanism, a term 
which implies a relatively simple neural effect of the task. He does 
not, however, make a distinction between boredom and fatigue, limit- 
ing himself to the purely ''objective" concept of work-decrement. 

The other factor tested by Robinson and Bills, conflict, is largely 
eliminated in the simple kinds of muscular action involved in experi- 
ments upon fatigue. Conflict refers to the effect of competition 
between different possible responses at various stages of a task. Bills 
gives the example of a subject who is asked to add and subtract 
alternately. Here the decrement is more rapid than it would be in 
continuous addition, although it is possible to become accustomed to 
such a procedure. 

While it may be useful for some purposes to state that these are 
general principles of work-decrement, at the same time it seems that 
this knowledge is of questionable value in increasing our understand- 
ing of the problems of fatigue. Let us note that Robinson and Bills 
reported that the effect of homogeneity was reduced if the subject 
became ''absent-minded." This is a parallel to the fact that habitua- 
tion reduces the influence of conflict. It may therefore mean that the 
rapid decrement is due to the fact that the subject is required to do 
something which is new and unnatural. Perhaps, then, both factors 
could be reduced to the well-known principle that new sequences of 
activity or new co-ordinations of movement are more fatiguing than 
activities which have become more habitual in their course. This 
argument could certainly be applied to the types of task which Robin- 
son and Bills employed in their experiments to demonstrate the roles 
of homogeneity and conflict. Whether it would also apply to other 
cases, such as the effect upon addition which Bills cites from Poffen- 
berger, might be argued. Perhaps in these other cases, however, 
there is a greater possibility of boredom's entering the picture. 

Output Decrement in Factory Work. — We see that decrement in 
performance with time does occur in laboratory studies of sedentary 
tasks, although it is variable and less predictable than the decrement 
in muscular tasks, and we are doubtful about how clearly it reflects 
diminishing capacity. Before we can evaluate output decrement as 



a useful index of fatigue, we must also consider the course of output 
through the working day in the normal working situation. 

Some of the factors which reduce the adequacy of the output 
indicator become more troublesome in studies of factory work. The 
motivational pattern is much more complex, especially if the worker 
is paid upon a piece-rate or bonus basis. Practice, however, should be 
less disturbing, since most workers would be at a stage where little 
improvement could be expected. 


















1sV2nd3r€4ih'5ttf 6fh7th' 8th 9th 10th 








Figure 7. Composite Hourly Performance Curves for Two Metalworking Plants 
(From Goldmark, et al., op. cit., p. 74.) 

Writers upon the subject of fatigue sometimes present a curve of 
hourly production which is considered to be "typical" of output in 
factory production. One basis for considering such a pattern as 
typical appears to be the average output curve reported in a study by 
the U. S. Public Health Service.^ This curve, shown at left in Fig- 
ure 7, is a composite derived from a large number of observations 
upon different workers and different kinds of factory work — all 
within the same automobile plant. This curve, however, results from 
canceling out various tendencies which appear in the curves of specific 
occupations or more homogeneous groups of occupations. In Figure 
8 we have reproduced several of these more specific curves from 

9 J. Goldmark, M. D. Hopkins, P. S. Florence, and F. S. Lee, Studies in 
industrial physiology: Fatigue in relation to working capacity: 1. Comparison 
of an eight-hour plant and a ten-hour plant, U. S. Public Health Service, Public 
Health Bulletin No. 106, 1920. 


[Ch. 4 

which the composite is derived. The "typical" pattern of decHning 
performance toward the end of the day appears most clearly in the 
heavier types of muscular work. In lighter tasks, it is by no means so 
clear, and sometimes it does not appear at all. 






































Figure 8. Composite Hourly Performance for Different Classes of Work (Eight- 
hour Metal working Plant) 

(From Goldmark, et al., op. cit., pp. 33, 45, 57, 67.) 

As the writers of the report point out, there are a number of 
factors which tend to mask any possible influence of declining capacity 
for performance. Among others, they mention the factor of stereo- 
typing. That is, the worker sets himself a constant daily quota. As 
an example, these authors mention that each of sixteen different 
workers drilled exactly 3600 pieces on every night of a given week. 
Variations in hourly rate are frequently a function of the worker's 
goal to maintain the daily average, rather than a function of fatigue. 

Ch. 4] 



Another finding of the same research group is that rhythmic opera- 
tions tend to be more constant in rate than do more irregular tasks. 

Another instance of the lack of correspondence between output 
and fatigue can be found in the curves of the same study of factory 
output. Figure 7 shows the composite curve already referred to for 
an eight-hour plant and another for a ten-hour plant, for essentially 
the same classes of work. If fatigue is responsible for decreased per- 
formance, and if fatigue is cumulative, it is difficult to see why the 
final decline occurs in the eighth hour in one case and does not ap- 
pear until the tenth hour in the other. In miscellaneous machine 
work, in the ten-hour plant, the high point of production for the 
afternoon came in the next to the last hour, that is, the ninth hour of 
the day. 

In practice, output curves may be of any shape ; many factors enter 
into their determination. In several instances even simple factory 
tasks showed improvement throughout the working days.^° This is 
interpreted as ''practice" improvement, although the data were not 
analyzed on a day-to-day basis to see whether the workers were ac- 
tually improving in the long run, or whether the improvement con- 
tinued to take place every day, with a loss over night. Practice is not 
the only possible interpretation of progressive improvements during 
the day, however. Link has interpreted the increase in output in in- 
specting small shells (Figure 9) as the result of the monetary incen- 
tives provided. 

Other investigators have found curves with a low point somewhere 
in the middle of the work-spell (morning and afternoon). This is 
interpreted as the result of severe monotony in a simple repetitive 
task.^^ We shall question the generality of this association in our 
later discussion of the topic of boredom. Nevertheless, here is an- 
other exception to the ''typical" curve. It would be difficult to argue 
that no fatigue is involved, simply because other factors have been so 
effective in producing an "abnormal" curve. If such reasoning were 
allowed, it would be possible to eliminate most "fatigue" by simply 
paying an extra bonus for keeping the production up during the later 
hours of the work-spell. 

Rothe has recently collected detailed data upon the output curves 
of butter wrappers, giving us another example of the difficulty of in- 

10 H. M. Vernon, Industrial Fatigue and Efficiency, New York, E. P. Button & 
Co., 1921, p. 14. 

11 H. C. Link, A practical study of industrial fatigue, Journal of Industrial 
Hygiene, 1919, 1, 233-237. 

125. Wyatt, J. A. Eraser, and F. G. L. Stock, The effects of monotony in work, 
Industrial Fatigue Research Board (Great Britain), 1929, No. 56. 



[Ch. 4 

terpreting these curves in terms of fatigue. Composite curves for 
groups of workers, and composite curves for the same worker over 
a period of time, were compiled. It was found that these individual 
curves may take any one of a number of different forms. The work- 
ing pattern of the same individual fluctuated widely from day to day, 
so that there was little predictability of the pattern for a given indi- 



7-d gJO ^^12 ^ ^.4 ^ 








Figure 9. Hourly Performance Curves for Visual Inspection of Cartridge Shells 
(From Link, op. cit., p. 236.) 

vidual. Some relationships did appear, however. There was a tend- 
ency for the patterns of different individuals to be similar for a given 
day. Composite curves for individual workers, but covering a period 
of several days, were interrelated. That is, two workers might show 
similar composite curves, even though an individual worker was not 
consistent with himself from day to day. Although these composite 
curves showed a low period of production toward the end of the after- 
noon, it is not stated how much production is reduced during this 

13 H. F. Rothe, Output rates among butter wrappers: I. Work curves and their 
stability, Journal of Applied Psychology, 1946, 30, 199-211. 

Ch. 4] 



time. If decrement does occur in this occupation, it is evident that it 
can be discovered only by studying average results from a large 
amount of data. 

Another possible index of the effects of fatigue would be a com- 
parison of the average hourly rate in the morning and afternoon. 
Vernon has collected extensive data upon this relationship. In his 
data, production is only 2 per cent lower in the afternoon, even on 
those jobs where the working day was ten hours long. 

These examples are enough to indicate the state of affairs. Per- 
haps many jobs do show a bona-fide fatigue decrement toward the end 
of the day (a decrement which is not due to decreased motivation or 
to the worker having completed a quota which he has set for himself). 
Even if we grant that possibility, there are sufficient exceptions to the 
"typical" pattern to make it a doubtful index of fatigue. If the typi- 
cal pattern appears, we still have difficulty in showing that the dec- 
rement is not a result of lowered motivation. If the typical pattern 
fails to appear, it cannot be taken as evidence of lack of fatigue. We 
can only say that the output criterion has failed as an indicator of 
variations in working capacity. 

Possible Approaches to a Study of Fatigue in Complex Tasks 

It is clear from the laboratory and factory studies of work dec- 
rement in complex tasks or ''mental" performance, that fatigue can 
be studied in this manner only for carefully designed tasks which 
may be quite unlike normal working activities. Even here the amount 
of decrement may be slight and difficult to measure reliably. There 
are at least four possible ways out of this difficulty : 

1. Limit experimental studies of fatigue to the laboratory, using 
only tasks which are known to show reliable decrement. Thus, if we 
wished to study the effect of nutrition, say, upon fatigue, we should 
measure its effect only upon those tasks which may be found to ex- 
hibit the property of rapid decrement {e.g., continuous addition, 
naming colors as rapidly as possible, canceling letters from a page of 
pied text). 

The obvious difficulty with this approach is that it assumes that all 
fatigue is alike, or at least that all ''mental" fatigue is alike. We 
should have to assume that a similar effect of a change of diet would 
be found for fatigability in other tasks, even though it cannot be 
measured. It overlooks the possibility, indeed almost the certainty, 
that the effects of work differ with the task. Consequently we should 



[Ch. 4 

not expect that a change in vitamin content would have the same effect 
upon the executive who is tired after a day of making responsible 
decisions as it would upon the automaton in the laboratory. At least 
we cannot assume, without investigation, that the effect would be the 
same. If we do not allow this assumption, however, the conclusions 
from such an experiment would have to be limited to such restricted 
statements that they would have little interest or practical value — 
e.g., ''Vitamin B has no effect upon fatigue in adding three-place 
numbers in columns of seven numbers for a period of four and a 
quarter hours." 

It is quite possible that the major sources of fatigue in many sed- 
entary and skilled types of work are not in the work itself so much 
as they are in the problems of social and emotional adjustment which 
the job entails. In the laboratory, these factors will operate in an 
entirely different manner from what they do in an office or factory. 
Laboratory studies would therefore give us an analysis of fatigue 
due primarily to the ''work itself." Without indices of fatigue which 
could be applied "in the field," it would be impossible to assess 
the effects of these complex factors of adjustment upon the net 

Of course it is important to know the effects of both sets of 
factors — the work itself, and the other elements of the total situa- 
tion. Laboratory studies, however, can give us only one side of the 
picture, unless it is possible to bring these emotional factors under 
laboratory control also. So far, this has not been achieved. 

If it is known that a certain group of activities have similar 
properties from the point of view of fatigue, e.g., a color-naming 
task and proofreading, and that one task is such that decrement is 
easily measurable, then the laboratory approach seems satisfactory 
and useful even if incomplete. This means that the use of the ap- 
proach now under consideration depends upon extensive investiga- 
tions of fatigue by other means which permit us to determine the 
relationships between the effects of different tasks. In other words, 
we must first show that color-naming in the laboratory has effects 
similar to those of proofreading in a publisher's office. Once these 
relationships are established, the method of decrement may be used 
as a short-cut procedure in solving a variety of practical problems. 

2. Where a specific task is not amenable to study by the decre- 
ment method, it may be possible to measure performance periodically 
in some other task which is more readily measurable. This is known 
as a fatigue test. Thus, in order to study the effects of the job of 



being a policeman, we might ask him to perform a series of tests in 
the morning, again before and after lunch, and finally as he leaves 
work. If his performance on these tests is diminished, we would 
assume that his job produces some kind of general fatigue which is 
reflected in other exacting activities. If some tests show greater 
effect than others, we would assume that there are specific effects of 
the work which are more limited in their nature — specific kinds of 

There are several major requirements for the use of the procedure. 
Most important, the tests must be sensitive to all important effects 
of the work. If not, a negative finding would be of no significance, 
since it might mean only that the tests missed the important effects 
of the work. This means, again, that the use of this procedure de- 
pends upon a fundamental analysis of the effects of the work by direct 
study, and that adequate fatigue tests come after this analysis rather 
than before. 

In practice, this requirement has not been met. Instead, fatigue 
tests have been based upon an empirical search for activities in which 
performance varies inversely with the duration of work. The in- 
vestigation is still in this stage, and has not yet reached the stage of 
inquiring whether these tests actually get at the important effects of 
the work. 

3. A third possible way out of this difficulty is a more refined 
analysis of those tasks which do not appear under gross measurement 
of output to show decrement. There is evidence that fine control of 
co-ordination, accuracy of skilled activities, timing, and the like may 
deteriorate, even though the effect of fatigue is not clearly evident in 
the over-all rate of performance or accomplishment. It might be, 
for example, that more accurate measurement would indicate "block- 
ing" or brief failures to respond even in such tasks. In skilled mus- 
cular movements, careful photographic records might reveal a slight 
decrease in accuracy of movement which is compensated for by in- 
creased effort. An investigation of techniques of this kind, which 
could be applied to a wide variety of tasks, would be of fundamental 
importance to the field of efficiency in human work. 

4. A fourth method of solution of the problem may lie in physio- 
logical indices of effort and fatigue. (See Chapter 5.) The ''invol- 
untary" bodily activities such as circulation, muscular tonus, and 
blood chemistry have received considerable attention as possible indi- 
cators of effort in mental work as well as in heavier muscular effort. 
Some of these techniques, though still in a trial stage, show definite 



[Ch. 4 

promise of validity. Whether they will also be convenient enough 
for field studies still is undetermined. If they are not, it will be neces- 
sary to perform much of the work in the laboratory, hoping to ex- 
trapolate the conclusions to the everyday working situation. 

Each of these approaches to the problem will be considered more 
thoroughly in the following sections, insofar as there is information 
available. Possibilities 2 and 4 have received more attention than the 
others, so that much of the discussion must of necessity be limited to 
them. This does not mean that these approaches necessarily show 
more promise. They have appeared more attractive to investigators 
so far, but they may receive relatively less attention as time goes on. 
The first approach mentioned is involved in many of the studies 
which will be described later when we take up such problems as dis- 
traction, the effects of alcohol, rest periods, and the like. There is 
little to be said now, however, upon the fundamental problems of the 
interrelations between fatigues developed under different conditions 
and with different tasks. We therefore pass directly to a more 
detailed consideration of the second approach. 

Tests of Fatigue 

In the search for measurable activities which would be sensitive to 
the effects of prolonged mental work, and skilled activities of all 
kinds, a great variety of tasks have been tried out. In order to inter- 
pret the results of these varied experiments, we must keep in mind 
the precautions which are necessary to show that a given test is a 
suitable technique. Let us outline some of the main ones. 

1. Statistical treatment of the results to determine whether the dif- 
ference between before-work and after-work test scores is significant. 

There are a number of factors which may lead to results which are so 
variable as to be indecisive: 

(a) The test itself may not be a reliable measuring device — e.g., 
it is not long enough to get a representative sample of perform- 
ance at the time. This is a difficult factor to control, because 
lengthening the test may make the test itself so fatiguing as to 
cover up the effects of the fatigue which we are attempting to 

(b) The test may measure a function which fluctuates with a 
variety of other factors. That is, the test may be an accurate 

1* A. G. Bills, Transfer of fatigue and identical elements, Journal of Experi- 
mental Psychology, 1932, 15, 23-36. 

Ch. 4] 



indicator of performance at the time of the test, but perform- 
ance itself fluctuates widely with other uncontrolled factors, 
(c) The use of a very small number of subjects, so that the results 
cannot be generalized beyond the particular subjects used in 
the experiments. If all the subjects consistently show the same 
trend, the number of subjects does not have to be very large. 
It should, however, be representative of a number of different 
sorts of individual — large, small; bright, dull; male, female; 
old, young ; etc. If the results for each individual are extensive 
enough to be significant for him, and if several individuals of 
widely different kinds show similar trends, then the results are 
of value without carrying out a mass experiment. 

N.B. If a test is to be used as a method of diagnosing fatigue 
in an individual, it must be shown to be significant for indi- 
vidual diagnosis as well as group differences. That is, a group 
of individuals might show a statistically significant trend if the 
group is very large, even though there are many individual 
exceptions and the changes in most individuals are very slight. 

2. Adequate control periods to eliminate the effects of practice, and 
of variations with the time of day, independent of working activity. 

For example, let us suppose that tested reaction time is lower after 
eight hours of office work than it was in the morning before work. We 
cannot conclude that this change is due to the work, even if it is not due 
to chance. The change might have occurred, even though the subject 
had not worked at all. The control tests taken on days when the subject 
does not work could determine whether this is possible. 

On the other hand, the performance might be higher at night than it 
is in the morning, as a result of practice. If so, it is necessary to measure 
and to allow for this effect in order to determine what effect the work 
had. Or, preferably, the test is repeated so many times that the practice 
effect is finally eliminated as a factor. 

3. Determination of the types of work for which a given test is 
applicable. For example, some tests might be affected primarily by pos- 
tural restrictions in the work, others by the visual aspects of the task, 
and so on.^^ This means sampling a variety of tasks as well as a variety 
of individuals, repeating the precautions described above for each task. 

The experimenters who have studied the problem of fatigue by 
such tests have given considerable attention to adequate control. As 

15 See T. A. Ryan, Varieties of fatigue, American Journal of Psychology, 1944, 
57, 565-569. 



[Ch. 4 

in the studies of muscular effort, however, there has been widespread 
neglect of proper statistical procedures. The reliability of the tests 
has often been completely unknown, or has been deduced from logical 
considerations rather than from a direct analysis. This would not 
have been fatal if there had been an over-all analysis of the statistical 
significance of the test variations in the experiment. Too often, how- 
ever, not even this was done. 

For example, Kraepelin, his students, and co-workers have pro- 
vided us with a number of studies of fatigue-tests in mental work. 
Yet some of these studies fail to report even the number of subjects 
employed or the number of observations on each subject, to say noth- 
ing of the data upon variability of the results. Conclusions are drawn 
upon the basis of simple averages. 

It is true, of course, that an experienced researcher in any field 
can tell by his direct observations whether or not the methods have 
great variability. He is still liable to error, however, and he should 
always check his impressions by statistical analysis of the data. If he 
does so, we assume that he will report the fact along w^th his results 
so that he can demonstrate clearly the value of his results. Unfortu- 
nately, however, many men w^ho early received their training in this 
field have failed to realize this need. 

Since so few of the experiments on fatigue tests so far reported 
have met the requirements laid down in the preceding discussion, we 
shall not devote much space to them except to indicate the general 
approaches that have been taken. The proposed tests may be 
roughly classified under the following headings : 

1. Tests of functional components. 

(a) Muscular strength and control (ergograph, tapping). 

(b) Sensory acuity and discrimination, e.g., discrimination of two 
points upon the skin. 

(c) Reaction time tests of varying degrees of complexity. 

(d) Attention and concentration. 

(e) Memory. 

2. Tests involving a variety of components (sometimes closely related 
to the fatiguing work itself). 

These cannot be classified readily, but a few examples would include 
these : addition or other calculation, writing to dictation, completing sen- 
tences, tests of speed of number checking. 

For more detailed accounts of these tests and results see M. Offner's Mental 
Fatigue, 1911, or Die geistige Ermiidung, 1928. Also, O. Graf's Uber Ermiidungs- 
messung durch Stichprohen, Kraepelin's Psychologische Arbeitcn, 1925, 8, 1-16. 



For many of these tests, the lack of statistical control of a single 
investigator has been brought to light by the attempts of other ex- 
perimenters to duplicate the results. Thus the literature is full of 
claims and counterclaims as to the value of a particular test. This 
could have been avoided if the precautions we mention had been 
taken. Offner, who summarized the work up to 1928, based his eval- 
uations to a great extent upon these contradictions.^^ From this 
rather haphazard collection of information, Offner concludes that 
muscular and simple sensory tests have little value in measuring the 
effects of ''mental" work. Greatest hope lies in the more complex 
types listed above under (2). 

Examples of Fatigue Test Studies. — Here are two examples of 
the use of the fatigue-test approach. The first was concerned with 
the effects of loss of sleep.^^ After observing her normal perform- 
ance in the tests to be described, the experimenter limited her sleep for 
three successive nights to 1^, 3^, and hours respectively. The 
tests were administered each day during the deprivation and for sev- 
eral days thereafter. Sometimes there were two sleepless periods 
separated by several normal days, with the test records complete for 
each day throughout the whole span. 

The tests were as follows : 

1. Dotting test. Small red circles on a moving tape had to be 
dotted as they passed. Circles of other colors had to be skipped, 
while the subject tapped a key with the other hand each time one 
of these circles of different color appeared. 

2. Association test. A series of words with logical connections be- 
tween them was read once. The subject tried to reproduce the 
list in the correct sequence immediately after it was read, 

3. Illusion of reversible perspective. This test involved a figure 
which was ambiguous so that it would be seen in either of two 
ways. The subject observed the figure for a period of time, 
signaling whenever the perspective was reversed. 

4. Speed of tapping. 

5. Learning and relearning of nonsense syllables. 

The most marked changes appeared in the first two of the tests. 
The performance was normal or above on days in which the loss of 

17 He also discards certain tests which he feels are too complex or difficult to 
administer to large groups, even if there is some hope that they may have value 
otherwise. This is due primarily to his interest in the use of fatigue tests in the 
schools for controlling the hours of work, scheduling, and the like. 

18 May Smith, A contribution to the study of fatigue, British Journal of 
Psychology, 1916, 8, 327-350. 



[Ch. 4 

sleep occurred.'^ It went into a decline after that, however, and 
sometimes remained below normal for more than a week. If further 
sleep was lost during this subnormal period, the decrease in sleep had 
the effect of returning the performance to normal. These results 
are, of course, only illustrative of the method. There was only one 

Figure 10. Test Performance Following Three Nights of Diminished Sleep (Dot- 
ting Test) 

♦After 1st vigil — 1^ hours of sleep. 
** After 2nd vigil — 3^ hours of sleep. 
*** After 3rd vigil — 5^ hours of sleep, 
(yigils were successive nights of diminished sleep.) 
(From Smith, op. cit., p, 333.) 

subject, SO that the value of the results is indeterminate, even though 
the changes were regular enough to be statistically significant for 
that subject. 

Another example is an experiment upon the effects of automobile 
driving. This is a more recent study, more carefully controlled. 

19 Smith, op. cit., pp. 333, 335. 

2 A. H. Ryan and Mary Warner, The effect of automobile driving on the 
reactions of the driver, American Journal of Psychology, 1936, 48, 403-421. 



The operators drove a distance of about 300 miles on ''driving days," 
which came every other day. On the ''control" days in between, they 
drove for an hour before the first test of the day, just as they did on 
the experimental days. The rest of the day was spent, however, in 
light activity and recreation. Six drivers were studied for a total of 


8 9 10 11 12 13 14 15 16 


*>fage for T 
preceding day 





Figure 11. Test Performance Following Three Nights of Diminished Sleep (Asso- 
ciated Words)" 

♦After 1st vigil — 1^ hours of sleep. 
** After 2nd vigil — 3^ hours of sleep. 
*** After 3rd vigil — hours of sleep. 
(Vigils were successive nights of diminished sleep.) 
(From Smith, op. cit., p. 335.) 

120 driving days. The results are given in terms of averages for the 
six drivers. 

The tests, which were repeated each day in the morning and again 
in the evening after driving or light activity, were as follows : 

1. Vascular skin reaction, a physiological test designed by A. H. 
Ryan, measuring the time of disappearance of a red streak on the 
skin produced by a stroke of a blunt instrument. 

2. Postural steadiness, the amount of sway in standing with the eyes 

3. Hand-eye co-ordination, inserting a metal stylus into holes of 
various sizes at a fixed rate, trying not to touch the edges. 



[Ch. 4 

4. ''Visual efficiency," Ferree's test, measuring the length of time 
a subject can maintain focus upon the letters //. The subject 
signals whenever the letters become blurred. (See Chapter 6 
for a discussion of this type of test.) 

5. Color naming, measuring the time required and the errors made 
in naming a series of 1200 color squares. A test which has been 
used in a number of special investigations — e.g., the effects of 
alcohol upon psychological functions. 

6. Mental addition. Adding columns of fifteen digits each. 

The results of the comparison of morning and evening tests are 
shown in Table 8, taken from Ryan and Warner's report. 

TABLE 8 * 

Change in Performance from Morning to Evening 


of Change 



Driving Days 

Control Davs 

P.E. Diff. 

Vascular Skin Reaction 

- 12.38 

- 1.21' 





Hand-eye Co-ordination : 

- 2.92 



Form Bf 


- 9.78 


Visual Efficiency 

- 2.62 

- .80 


Color Naming: Time . 


- 1.65 


Color Naming: Errors 


- 16.28 


Mental Addition 


- 3.59 


* Modified from Ryan and Warner, op. cit., p. 409. 

** A ratio of difference to probable error of difference should usually be 4 or greater if 
the difference is to be considered significant. Values slightly smaller are sometimes con- 
sidered as suggestive. 

t Form B of the co-ordination test was longer and more difficult than Form A. 

What conclusions can be drawn from a study of this kind? Ryan 
and Warner, authors of the report, draw the following : 

"(1) There is a demonstrable fatigue effect of a long automobile 
drive on body reactions. 

(2) Long automobile drives tend to decrease the fading time of 
the vascular skin reaction; increase unsteadiness in standing; 
decrease the accuracy of hand-eye co-ordination ; decrease the 
visual efficiency; decrease the speed and accuracy of mental 

(3) Greater variability occurs within the test scores made after 
long automobile drives than before them in postural steadi- 
ness, hand-eye co-ordination, and mental addition. 

Ch. 4] 



(4) The tendency of long automobile drives is to produce a loss 
of effectiveness of certain sensory discriminations, association 
processes, and motor reactions similar to those required in 
driving. These observations suggest that the effect of a long 
automobile drive may render a driver temporarily prone to 

These conclusions are cautious, and they appear justified by the 
results we have reported. Since accident liability in performance 
cannot be easily measured, it seems necessary to study other functions 
w^hich lend themselves more readily to reliable measurement. 

Since, how^ever, w^e are concerned v^ith the evaluation of this 
general approach to the measurement of fatigue, must examine 
the meaning of this study more fully. We must raise some questions 
which refer not only to the particular problem of this experiment, 
but also to any other problems which might be approached with the 
same general method of fatigue tests. 

TABLE 9 * 

Proportion of Days Showing Decreased Performance 


Driving Days 


Control Days 


Vascular Skin Reaction , 









Pland-eye Co-ordination : 

A .... 54 




B .... 65 



Visual Efficiency 





Color Naming: Time , 





Color Naming: Errors 







Assembled from data in Ryan and Warner, op. cit. 

1. In order to obtain reliable measurements, the time spent upon 
each subject was so great that only six subjects could be studied. 
Thus, while the results are evidently significant for the group, there 
is still the question of how the driving would affect other groups of 
individuals. Since no results for single individuals are given, we 
cannot tell how much of the variability of remits is due to the varia- 
tions of each individual and how much to differences between indi- 
viduals independent of their temporary variations. If each individual 
was consistent but there was wide variation between individuals, 
there is a definite need to study more subjects. Table 9 shows some 

21 Ryan and Warner, op. cit., 420-421. 



[Ch. 4 

data which Ryan and Warner give as an indication of the consistency 
of the results, but these values are also composite results for several 

2. The decision that these tests represent functions which are 
involved in accident liability is based upon judgment, and it is doubt- 
ful whether it could be directly proved. Such judgments are very 
difficult to make, as we shall see when we discuss accident control. 
(See Chapter 12.) They must be made with great caution. 

3. How are we to measure the amount of fatigue produced by the 
driving? A number of important practical applications would de- 
pend upon the answer to this question. How many hours of driving 
can be performed safely before the changes in function become great 
enough to be dangerous? Suppose that a new car design is to be 
tested for ease of driving. Can we differentiate between the effects 
of driving two different cars by a procedure like that used by Ryan 
and Warner ? 

All we know is that the changes reported are significant statisti- 
cally. How important they are for practical driving is not determined. 
Also, it is doubtful if the tests are sensitive enough to distinguish 
between the effects of driving under two different conditions. At 
least these are questions which are still to be determined. 

It is likely that further research into the problem of fatigue tests 
will be valuable in throwing more light upon the interrelations of 
the effects of various kinds of work. That they will provide us with 
a method of measuring or even estimating the cost of work at any 
time in the near future is highly doubtful, for the reasons which we 
discussed in evaluating A. H. Ryan's study. 

There is probably much more to be gained from this type of re- 
search if it is made more systematic, and if it is used as a device for 
finding out the nature of the effects of work (fatigue). Those fa- 
tigue tests which have measured so-called ''elementary" functions 
have been based upon extremely vague hypotheses that fatigue affects 
''attention," "speed of reaction," or similar broad aspects of func- 
tioning. It has then been assumed that the tests employed are valid 
measures of these functional characteristics. Those who have re- 
garded the fatigue test 'as a measure of elements or parts of the job 
itself under controlled conditions (as in A. H. Ryan's study of 
drivers) have not been concerned with problems of the general effects 
of fatigue. In fact, studies of this latter kind are simply substitutes 

2 2 See H. M. Johnson, Index-numerology and measures of impairment, American 
Journal of Psychology^ 1943, 56, 551-558. 

Ch. 4] 



for direct measurement of work decrement. They gain, in that the 
tests are more amenable to quantification than the task itself, but they 
lose, in that the validity of the tests as samples of performance in 
the task itself is unknown and usually must be assumed. 

These studies are and will be more enlightening if we take them 
out of the context of measurement of fatigue and fatigue "tests" 
altogether. If, instead, we ask what kinds of activity are functionally 
so related that work in one task produces decrement in the others, 
we are entering the context of the general problem of fatigue transfer 
rather than of fatigue measurement. More information of this kind, 
on a wider variety of activities, could lead to the next step, which 
would be the formulation of general principles to cover these func- 
tional relationships. We might then be on our way to discovery of 
just what it is which brings about the functional relationships between 
certain tasks — what bodily resources they draw upon in common, 
what common elements there are in a group of related activities, and 
so on." Such a program calls also for the use of many physiological 
techniques that are to be discussed in the next section. 

Some attempts have already been made in this direction, but they 
have been based upon such scanty information that the principles 
enunciated have been, of necessity, exceedingly vague and trite. 

Bills, for example, summarizes the conditions ''under which the 
decrement developed in one task can be expected to transfer to a 
second task which is engaged in directly afterward. These con- 
ditions are : 

When the tasks are both mental, or both physical, or when both have 

elements of each. 
When the two tasks are quite similar, involving the use of some of 

the same neuromuscular processes. 
When the work set, i.e., the directing and controlling attentive set, 

is common to both. 
When the first task is so complex that it results in the widespread 

fatigue of many mechanisms. 
When the setting in which the work is done remains unchanged, 

provided this feature was the fatiguing element in the first task." 

While this conclusion of Bills is all that is justified by our present 
knowledge of the transfer of fatigue, it is obvious that it does not 

2 3 As an example of this kind of analysis applied to visual work, see M. E. 
Bitterman, Transfer of decrement in ocular tasks, American Journal of Psychology, 
1946, 59, 422-438. 

24 Bills, op. cit., p. 111. 



[Ch. 4 

carry us very far toward an understanding of fatigue. Of course, 
fatigue-transfer is a problem of interest in itself, but it has even 
greater potential importance if it can be used to throw light upon the 
underlying bodily events responsible for fatigue and for transfer of 

A fact which introduces some confusion into the field of fatigue 
testing is that sometimes work produces an improvement in test scores 
rather than an impairment. If the improvement is really due to the 
work and not to some other factor, it may be interpreted in several 
ways. First, the effects of a specific kind of fatigue may be to reduce 
capacity for some tasks (closely related to the fatiguing task) while 
increasing the capacity for unrelated tasks. It might be simply that 
the task provides a change or relief from the restraints imposed by the 
fatiguing work. Another interpretation is suggested by Johnson, 
who compares fatigue with the effects of alcoholic intoxication. The 
apparent improvement may be only temporary, and may not repre- 
sent a real benefit to the organism. Instead, it may indicate a 
condition which is ultimately harmful because it makes additional 
demands upon the reserves of the body. 

Perhaps the best way to approach this paradoxical improvement 
resulting from work is to wait until we learn what kinds of bodily 
change underlie the effect. 'Tmprovement" and ^'impairment" are 
values arbitrarily imposed upon the test scores in relation to their 
involvement in practical affairs. It is as if we were to call an in- 
crease in heart rate an ''improvement," and a decrease an "impair- 
ment." Obviously, we are justified only in saying that the heart rate 
has changed. In the same way, a change in test performance is to be 
regarded as a result of complex internal bodily changes. Whether or 
not these changes represent important impairment of bodily economy 
still remains to be determined. 

So far, what has happened is that those tests which fail to show 
decrement are discarded as failing to reflect the effects of fatigue. As 
a result, those tests which show improvement are given the same 
treatment as those tests which show no effect at all. If, as suggested 
above, the tests were used to examine the various effects of work 
without pointing so specifically toward developing indices or measures 
of fatigue, any effects of work would be important, no matter what 
direction of effect happens to be involved. The task of the scientist 
is, then, to piece all this information together so as to formulate 
hypotheses concerning the underlying bodily changes. 

2 5 H. M. Johnson, The real meaning of fatigue, Harper's Magazine , 1929, Vol. 
158, 186-193. 

Ch. 4] 



Change of Capacity as Indicating Change in Efficiency. — Another 
way in which tests of capacity are often used in the study of efficiency 
differs from the approach we have spoken of as the "fatigue test." 
In investigating the effects of drugs, changes in atmospheric condi- 
tions, or other factors, the experimenter may measure the psychologi- 
cal effects by comparing test performance with and without the drug, 
or under varying temperatures, oxygen pressures, or the like. If it is 
found that alcohol, for example, impairs ability to type, it is usually 
said that alcohol impairs "efficiency" in this performance. 

In some cases, this mode of expression may be no more than a 
confusion of output alone with efficiency, a usage which we criticized 
at the outset of this book. Frequently, however, efficiency as a ratio 
of output and cost seems to be implied. We must therefore consider 
whether this conclusion about efficiency on the basis of impaired 
capacity is justified. 

Let us accept, at least for purposes of argument, the position that 
the tests of reaction time, number checking, color naming, speed of 
adding or learning, which are used in these experiments, do measure 
capacity. At least the subjects are working under high motivation, 
since they are usually co-operative and told to do their best. 

If we grant that the subjects are working at their limit of perform- 
ance in the test, and that the drug under study is found to produce a 
reduction in this limit, we still have to relate this finding to the con- 
cept of efficiency. The argument would have to be something as 
follows : 

1. As the organism approaches its limits of performance, it becomes 
less efficient. For example, a sprinter at top speed uses more 
oxygen per mile than he does at an easy lope or at a brisk walk. 

2. If the worker's limit of performance (capacity) is reduced by 
some factor, such as a drug, his usual rate of performance on his 
job will be closer to the limit than it normally is. 

3. Although his rate of performance in his normal tasks does not 
change, he is now exerting greater effort, just as though he had 
speeded up in his rate of work. (See definition of effort in 
Chapter 2, page 22.) 

4. Therefore the drug has not only reduced the maximum limits 
of performance, but it has also reduced the efficiency with which 
the task can be carried on at normal rates. 

Notice the assumptions involved. First it is assumed that these 
tests measure capacity, or at least that the change under the influence 
of the drug indicates a change in capacity. Then it is assumed that 



[Ch. 4 

efficiency of performance at a given rate is a function, not of the rate 
itself, but of the relation of the rate to maximum possible perform- 
ance (effort). In other words, we are assuming that other elements 
of cost increase in proportion to the effort involved. 

Another way in which we might justify using these results as 
indices of efficiency would be as follows : It is assumed that these 
tests measure aspects of alertness, speed of reaction, and attention, 
all of which are necessary for accurate performance of complex and 
dangerous jobs such as automobile driving or machine operation. If 
these capacities are impaired, the danger of serious error and accident 
is increased, an obvious source of inefficiency. 

At present this latter set of assumptions is not so well founded as 
the first. There is no evidence for the validity of the tests commonly 
used in experimental studies of drugs or physical factors as measures 
of the kind of alertness or reaction speed involved in dangerous occu- 
pations. From what is known, the writer would prefer the former 
approach in terms of the relationship of performance to capacity. 

Let us return now to the assumption which we accepted for pur- 
poses of argument, namely, that test performance reflects changes in 
capacity. Essentially the same question arises in any measure of the 
effects of some factor upon performance. Thus the measurement of 
fatigue in terms of output decrement assumes that decreased capacity, 
due to fatigue, is reflected in reduced output. The measurement of 
performance in a fatigue "test" before and after work is based upon 
a similar assumption. As we have said (see Chapter 2), capacity is 
the maximum performance of which a subject is capable under given 
conditions. Even though the motivation of the subjects of these ex- 
periments is high, it is probably not maximal. In fact, the term 
''capacity" refers to a limiting value which is never reached in prac- 
tice ; it is only approached. 

The crucial question, however, is not whether test performance 
is nearly maximal, but whether the changes in performance which oc- 
cur under experimental variation accurately reflect the changing 
limits. There is some evidence that even under laboratory conditions 
the worker has enough reserve so that when conditions become ad- 
verse he exerts greater effort, and thus partially or wholly obliterates 
the effect upon performance which we should expect to appear. (This 
statement is based upon indirect evidence, or upon physiological 
measures to be discussed on subsequent pages.) 

Thus, even under carefully controlled conditions with highly mo- 
tivated subjects, changes in test performance are only partially a 
reflection of changes in capacity. It is probable that a reduced per- 



formance under these conditions does indicate a reduced capacity. A 
failure to observe decreased performance would not, however, be a 
conclusive demonstration that the capacity remains the same. 

Summary of Fatigue as Indicated by Decrement and Fatigue 
Tests.— Our analysis of the studies of output decrement in sedentary 
activities has led us to the conclusion that this index has little value 
as a reflection of fatigue in ordinary daily activity. Outside of the 
laboratory there are many factors which prevent any change in ca- 
pacity of the worker from showing its effect in a decline of actual 
productive rate. Even in the laboratory it is only specially selected 
tasks which show clear-cut and reliable decrement which is a function 
of the length of working period. 

As in the case of muscular work, then, it is possible to study output 
decrement in the laboratory if the experimenter can limit himself to 
certain narrow restrictions in the experimental task and in the condi- 
tions of work. In most cases, however, these restrictions mean that 
the conclusions apply to such an unusual kind of task that the applica- 
tion of these conclusions outside of the laboratory is of little value. 
Only if the laboratory task respresents a phase or component of some 
more complex normal task is this approach likely to be useful. 

Among the other possible approaches to the study of fatigue, one 
other general method has been given extended consideration in this 
chapter. This is the procedure called the ''fatigue test." We have 
described some of the normal research precautions which must be 
applied before the results of such a study can be interpreted at all. 
This discussion of methodology was essential because it has been 
neglected in so many of the published researches in this field. The 
result is that only a few researches upon "fatigue tests," can be given 
serious consideration. The remainder contribute only results which 
may be spurious because of the unreliability of the tests, insufficiency 
of sampling, or uncontrolled variables. 

When the ordinary experimental and statistical precautions have 
been taken, the approach by way of fatigue tests is a way of gathering 
valuable information upon the effects of work. The general scientific 
program based upon fatigue tests would be to determine the interre- 
lations of a great variety of tasks from the point of view of the trans- 
fer of fatigue-effects. Such a program would be directed toward 
finding the underlying effects of work by a study of the pattern of 
interrelationships brought out from fatigue tests. 

The use of fatigue tests for the solution of technological problems 
requires other steps beyond the general scientific program sketched in 



[Ch. 4 

the preceding paragraph. In evaluating the results of sample fatigue- 
testing researches, we have found certain deficiencies in the results 
from the point of view of application. Since we treated these diffi- 
culties in piecemeal fashion, as the problems arose in discussing our 
examples, we must now summarize what is necessary in order to 
make fatigue tests practically useful. 

1. Most studies of fatigue tests have been content with showing that 
work produces a significant change in test performance — a change which 
is significantly greater than those which appear in the absence of work. 
We must also show that the test performance varies with the amount of 
work. In other words, we wish to show that the test is sensitive to 
varying amounts of effort, and the only way open (in the absence of a 
direct measure of effort) is to vary the total amount of preceding work 
over a considerable range. 

If a test has met this requirement, it will be useful for certain prob- 
lems of machine design or work-methods. In Ryan and Warner's meth- 
ods, for example, if we could show that six hours of driving car A is the 
equivalent of eight hours of driving car B, then we should have weighty 
evidence in favor of the design of car A. Until fatigue tests are made 
sensitive enough to produce results like this, however, the method re- 
mains of largely theoretical importance. 

2. It is also necessary to evaluate the changes measured by the fatigue 
test in terms of their meaning for the total economy of the individual. 
The fact that car driving slows verbal reaction time (color naming), for 
example, is meaningless in itself. To evaluate this change in speed as a 
deterioration of the individual is to assume either (1) that slowing of 
reaction means that the individual is less capable of performing his job 
or the other tasks which he is expected to carry out, or (2) the slowing 
of reaction is a reflection of some physiological change which is harmful 
to the individual if it exceeds certain limits. Neither of these assump- 
tions has any real justification in terms of experimental evidence for the 
reaction time test or for any of the other tests in use in the studies we 
have sampled. Of course, it seems only ''common sense" to make the 
assumptions. As we have stressed before, however, this does not make 
the assumptions necessarily correct, and experimental justification is 
still needed. 

The program of fatigue-testing for purposes of application to indus- 
trial problems must therefore be a very extensive one. It is not only 
necessary to investigate the effects of work upon the tests, but we must 

Ch. 4] 



know how performance on these tests is related to other criteria of the 
capabiHty, proneness to accident, health, or general satisfaction of the 

3. For some purposes, fatigue tests must be applied upon an indi- 
vidual basis. In these cases, the tests which are to be applied must meet 
further requirements. It must be shown that the test is significantly 
affected by work in the single individual, as well as in the large group 
of subjects. It would be quite possible that a test could show a signi- 
ficant average effect in a large group of subjects, even though some of 
the subjects consistently showed the reverse effect or no effect at all. A 
test which did not show such wide individual variations would be valuable 
for both individual and group studies, since variability in the latter case 
would be reduced by the elimination of the extreme individual variation. 

4. A program of fatigue-testing must sample a wide variety of tasks 
and activities — as nearly a complete sampling of human activities as pos- 
sible. To return to an example used once before, to show that driving 
car A has a greater effect upon a battery of fatigue tests than that pro- 
duced by car B, is not conclusive unless we rule out the possibility that 
another battery of tests could show a reversed result. This requirement 
means that we must have a wide sample of test activities, as well as a 
sufficient sample of activity upon each test. Once it is shown that the 
results for certain groups of tests are consistent with one another, we can 
save time by using only one sample from the related group of tests. Be- 
fore this can be done, however, we must again await the results of more 
extended research. 

It is thus evident that, although fatigue-testing is a field of re- 
search which must be carried on for some time, it is of a sort which 
lays the groundwork for later application. The applications can be 
made only when and if the above characteristics of fatigue tests can 
be demonstrated, which means only after much intensive additional 

The literature upon fatigue-testing in recent years gives the im- 
pression that interest in the problem has been declining. Perhaps this 
is due to discouragement at the vastness of the problem, or to the 
belief that so much has already been attempted with so few tangible 
results. Yet the amount of satisfactorily controlled work has been 
relatively small in relation to the size of the problem. Perhaps prog- 
ress can be accelerated if new procedures of testing, not now en- 
visaged, can be discovered. 



[Ch. 4 

In all tests based upon psychological performance, however, there 
will always be a difficulty of controlling the motivation of the subjects 
taking the tests. For this reason, as well as for others, many re- 
searchers have preferred to search for indices of fatigue and effort 
which require only the measurement of physiological changes. These 
techniques will be the subject of the following chapter. 

Chapter 5 


Since both sedentary and muscular work entail the co-operative 
functioning of many parts of the organism, it is reasonable to expect 
that sedentary work would be accompanied by various measurable 
physiological changes, and that fatigue resulting from sedentary 
work would have a physiological basis. A large body of experimental 
data has therefore been assembled in the search for physiological cor- 
relates of mental effort and the fatigue resulting from mental work. 

Although we have already given some preliminary consideration to 
the concept of effort in an introductory chapter (Chapter 2), it should 
be enlightening to consider this concept more thoroughly at this 
point. If we seek correlates and indices of effort, we must first have 
a clear notion of what effort is. 

An Analysis of the Concept of Effort 

Although the term ejfort plays a crucial part in many phases of 
applied psychology and industrial engineering, it is a term whose 
meaning has been very inadequately defined. Time standards for 
jobs are supposed to be determined in such a way that workers of 
equal training and experience would receive equal pay for equal ef- 
fort. Methods-design is the problem of finding less effortful methods 
of performing a job. The term is used in common-sense analyses of 
these problems without definition, and a need for definition is not 
felt because it refers to a meaningful aspect of common experience. 

The experimental study of efficiency, however, requires that we 
deal with definable and, if possible, quantifiable phenomena. Often, 
terms from common speech cover too many distinct and unrelated 
phenomena to be scientifically useful, and such terms must be more 
carefully limited in their use if they are to be valuable to the research 
worker. We must therefore consider the various meanings of the 
term, in order to find that meaning which will make it most readily 




[Ch. 5 

amenable to experimental study, and, more important, which will 
delimit a unique and important area for investigation. 

The possibilities which we shall seek to weigh are as follows : 

(a) Effort is equivalent to energy consumption. 

(b) Effort is a general term covering a number of separate elements 
of the cost of work. 

(c) Effort is a term describing certain aspects of psychological 
functioning — ^the pushing on toward completion of a task. It 
is a term describing the ''experience" of the worker as he per- 
forms his job. 

(d) Effort refers to the rate of performance of an individual in 
relation to the maximum possible rate of performance at the 
time and under the given conditions. 

Each of these notions of effort has some sanction in common 
usage, and each leads to different consequences in practice. The first 
two meanings would imply that eft'ort is simply a synonym for other 
aspects of the cost of work, so that the word could be used for variety, 
or for brevity, as in the case of (b). Unfortunately, common sense 
implies that there is more involved, and we must investigate this 
possibility that effort concerns an aspect of activity not covered by 
other terms. We do not desire to overlook any possibilities even if 
they are complicated or inconvenient. 

(a) Certainly, effort is more than energy in the strictly physical 
usage of the latter term. As we have seen, mental work which the 
worker would describe as extremely effortful does not show a cor- 
responding increase in metabolic activity in the organism. In refer- 
ence to possibility (&), all the elements of cost so far considered, such 
as fatigue, dissatisfaction, loss of health, risk of accident, refer to 
effects of the work upon the individual, for the most part, aftereffects. 
Effort, however, is supposed to refer to some aspect of the activity as 
it progresses — it is a characteristic of the work-in-course. ThesQ 
other elements of the cost of work cannot, therefore, be identified 
with effort, although we may expect a close correlation between total 
cost, as measured by them, and the level and duration of effort, how- 
ever effort is defined. 

(c) The third possible meaning would be stigmatized as "intro- 
spective" by those psychologists who insist upon being "objective" at 
all costs. The present writer would have no prejudice against such 
an approach if it could be made workable. In fact, we may eventually 
be forced to adopt this approach to effort, since it is the closest to the 

Ch. 5] 



common-sense meaning, and therefore is intimately involved in the 
practical problems which are placed before the psychologist in in- 
dustry. That is, if the worker or the industrialist seeks to have pay 
awarded on the basis of ''equal pay for equal effort," the thing which 
is sought is only indirectly indicated by fatigue and loss of health. 
What interests the practical man is that a person who is ''working 
hard," "who is exerting himself," or "taking pains," should be paid 

If fatigue or other subsequent indicators were easily and accurately 
measurable, then the problem might be solved as a matter of expedi- 
ency by equating effort with these effects or correlates of effort. As 
we have seen in the preceding chapters, however, fatigue is a difficult 
problem in itself. Perhaps suitable direct criteria of effort would be 
more practically useful than measures of fatigue. Certainly they 
would be more immediately measurable if they could be discovered, 
since we would have a measure which could be applied during the 
course of the work itself. 

Effort could be defined in terms of Bentley's functional approach 
to psychological description.^ Lacking experimental studies directed 
toward obtaining refined descriptions of the effortful aspects of ac- 
tivity, we are limited to conjectures as to the probable nature of this 
description.^ Effort obviously refers to the determined or searching 
aspects of function. 

Such a descriptive study should ultimately be carried out, both 
because it would contribute to our fundamental understanding of the 
organism at work, and also because it might suggest new modes of 
approach to the practical problems of measuring and controlling ef- 
fort. At present, however, one cannot see what such a description 
would contribute. Moreover, the descriptive approach still leaves 
open the problem of measurement, which is so important in the solu- 
tion of industrial problems. 

(d) Although definite advances in the status of our problem may 
come from the descriptive approach just described, nevertheless we 
need some simpler and more definite statement of what is meant by 
effort. We have found the definition given above under (d) a useful 
way of conceiving effort, since it is in terms which are at least ap- 
proximately measurable, and also remains fairly close to the common- 

1 M. Bentley, The Neiv Field of Psychology, New York, D. Appleton-Century 
Co., Inc., 1934. 

2 For illustrations of the technique employed in such a functional description 
see the series of studies published as the Cornell Studies in Dynasomatic Psychol- 
ogy, under Bentley's editorship, American Journal of Psychology, 1938, 51, 203-360. 


sense meaning of the term. It would appear to be the meaning which 
is implicitly involved in many of the physiological studies of mental 
work which we are now about to examine. 

In spite of our adoption of definition (d) as being the most man- 
ageable statement of the meaning of effort, it must be recognized 
that there are definite flaws involved. One of these is the practical 
difficulty of measuring capacity as of a given moment. This difficulty 
can probably be overcome by statistical and experimental controls. 
Another flaw is inherent in the use of a rate of output as a measure. 
Suppose, for example, that the capacity of a given individual at a 
given time is 100 units per hour. He increases his rate of work from 
50 to 60 units per hour, an increase of 10 percentage units of effort. 
Later, he increases his rate once more from 60 to 70 units per hour, 
or by another equal step upon the arbitrary scale we have set up. 
Common observation, however, would make us suspect that the 
second increase is more difficult than the first, that the change in 
effort is greater in the second step than in the first. This would be 
an especially important factor when the performance varies close to 
the upper limit. 

Lacking any way of measuring effort in anything except output 
terms, however, we must be content with it for the time being. We 
must be cautious, though, about assuming that effort is to be defined 
as a linear function of the ratio of performance to capacity. More 
complete knowledge may lead us to adopt a definite functional re- 
lationship. At present we must be content with the relationships 
"more" and "less." This is at least a first step toward measurement. 
In later discussions in this book we shall suggest researches which 
are needed in order to throw more light upon effort and its relation 
to other factors in work. 

The extended discussion of tests of fatigue involving psychological 
activities included many considerations which apply equally well to 
physiological measures of effort and fatigue. Essentially the same 
requirements of experimental and statistical control apply to these 
physiological tests, so it will be necessary to discuss only the extent 
to which these requirements have been met. 

A variety of physiological changes in the state of the organism 
during and after prolonged mental work have been reported. Changes 
in pulse rate and blood pressure, breathing, the vascular state of the 
skin (cf. A. H. Ryan's test mentioned above), changes in cell bodies 
and chemistry of the blood, urinary changes, muscular tonus or 
tension, have all been reported by some investigators to be correlated 
with the length or difficulty of mental work. Of these, the major 

Ch. 5] 



emphasis has been placed upon circulatory measures (pulse and blood 
pressure) and measures of muscular tension. 

Most of these functions are known to be closely related to heavy 
muscular effort. They are also activities of bodily systems which 
are involved in emotional seizures. It has been argued that they 
should therefore be related to effort in general, and that they might 
serve as measures of effort even though metabolic indices are unre- 
liable when applied to mental tasks. 

It has also been claimed that ''thinking is muscular work" ^ be- 
cause all psychological activities seem to be accompanied by muscu- 
lar contractions, and also because changes in muscle tension may 
facilitate or hinder psychological performance. The bodily phenom- 
ena accompanying tasks involving alertness and attention should 
therefore reflect patterns similar to those involved in muscular work. 
Such a priori arguments are, however, inconclusive for several 

In the first place, the argument that thinking is muscular work, 
and that attention is muscular tension, goes far beyond the demon- 
strated facts. It is true that we usually find muscular tension in- 
creasing when the subject is alert, when he is solving a problem in 
mental arithmetic, and so on. This does not justify equating the two 
types of activity, and even though we would grant that tension is an 
element in all activity, ''mental" as well as muscular, mental work is 
certainly much more than a pattern of taut muscles. Even in work 
which is classified as primarily muscular, it has not been demonstrated 
that all fatigue is localized in the muscular apparatus. Particularly 
in skilled muscular activities, it is probable that effort is involved in 
the control of the muscular patterns as well as in the contractions 

The above objection still admits a role of muscular tension in 
mental work. In fact, we shall attribute considerable importance to 
this role, in a somewhat different way. There is, however, still 
another flaw in the a priori argument that physiological patterns 
should be similar in the whole range of mental and muscular work. 
In those activities in which the muscular activity is not central and 
overt — e.g., mental arithmetic, composing a report, and the like — the 
muscular tensions are present but they are so slight as to call for 
relatively little energy. If they were not, this energy could be clearly 
demonstrated by the metabolic technique. 

3 Bills, op. cit., pp. 10-13. See also, A. G. Bills, Fatigue in mental work, 
Physiological Review, 1937, 17, 436-453. 



[Ch. 5 

How, then, can it be argued that fatigue in mental work is similar 
to the fatigue due to heavy muscular labor? If it were, we should be 
able to continue our efforts in sedentary tasks almost indefinitely 
without appreciable fatigue, so long as we changed our position often 
enough to avoid muscular stiffness. 

Perhaps some will say that the "real" fatigue in mental work is 
relatively slight or non-existent; that the fatigue which is reported 
after several hours of work upon a difficult problem is only "subjec- 
tive." Some authors, indeed, seem to imply that reported fatigue is 
a sort of unreal existence without organic basis. "Subjective" fa- 
tigue must have some bodily basis, however, and the finding that it is 
not the same basis as muscular fatigue would not make it belong to 
the realm of "purely mental existence" (whatever that may be). 

Others may say that the primary source of fatigue in mental tasks 
is due to emotional strain rather than the effort involved in just pay- 
ing attention. This is very probable, but the fatigue resulting from 
emotional seizure cannot be accounted for on the basis of muscular 
tensions either. These tensions are likewise not great enough to 
explain the severe effects of a prolonged emotional upset purely in 
terms of the energy consumed. 

Naturally, the value of physiological indices of circulation or ten- 
sion can be settled only by direct experimentation. The a (priori 
arguments analyzed above are mentioned because they are so fre- 
quently brought up in discussions of fatigue, not because they con- 
tribute to the solution of the problem. In fact they are more likely 
to lead us astray in any search for a solution. 

The factual evidence on the effect of various kinds of sedentary 
work upon physiological functions is very similar to the evidence 
upon psychological tests of fatigue. There is the same tendency to 
report results of a small group of subjects, without a statistical analy- 
sis of the data which are reported. Again the result is a series of 
conflicting reports upon the same physiological indicator. For ex- 
ample, the pulse rate is variously reported as increasing, decreasing, 
or showing no change while a subject works upon numerical compu- 
tation or similar tasks. 

Circulatory Indices. — Bills concludes that all work involves an in- 
creased pulse rate, but that sedentary work may appear to reduce the 
rate if it follows soon after walking or other overt muscular activity.* 
This is because the level of circulatory activity in sedentary work is 

4 A. G. Bills, Fatigue in mental work, Physiological Review, 1937, 17, 436-453. 

Ch. 5] 



lower than that required in muscular activity. In order to demon- 
strate clearly the effects upon circulation, it is necessary to establish 
a base line for the subject by an extended rest before the work begins. 
This is true, of course, of any physiological indicator which may be 

If the work is done calmly with little emotional strain, pulse, blood 
pressure, and other circulatory indices seem to reflect the relative 
amount of muscular activity going on in the body. As such, they can 
be used as rough measures of metabolic cost of work. They are 
somewhat easier to apply than direct measures of metabolism, how- 
ever, and for that reason may be more useful under many conditions. 
Special emotional strain produced by the working situation will also 
be reflected in these circulatory measures, unless the subject was also 
upset during the period of rest which established the base line. If 
the situation is such that it produces an unfavorable anticipation 
before the work begins, there is no way of establishing a base line 
and no way of estimating circulatory increases. 

It is probable that, under carefully controlled conditions, work 
such as problem solving or mental calculation and the like will pro- 
duce significant increases in pulse rate and blood pressure over the 
resting values. 

Gillespie, for example, found that both pulse rate and blood pres- 
sure were higher in subjects doing mental calculation than they were 
when the subjects merely read the numbers.^ No statistical treat- 
ment is given, however, and there were some exceptions to the general 
trends. He concludes that the rise in blood pressure is not propor- 
tional to the difficulty of the work. 

Whether different levels of effort can be distinguished, and how 
valuable these tests may be for measurements in the field under 
poorly controlled conditions, remains to be discovered. Pulse rate is, 
of course, more easily measured than blood pressure, and it can be 
recorded continuously during work by means of the cardiotachometer 
or other devices. We can accept the statement that the pulse rate is 
increased by mental work in comparison to a resting value obtained 
when the subject is relaxed and calm. The changes may be small, 
however, and diflicult to observe except when under carefully con- 
trolled conditions and with large numbers of observations. Then 
too, the establishment of the resting value is difficult because of the 
excitement which may come with the experiment. In order to con- 

6 R. D. Gillespie, The relative influence of mental and muscular work on the 
pulse rate and blood pressure, Journal of Physiology, 1923-1924, 58, 425-432. 



[Ch. 5 

trol the latter it is necessary to work with subjects who are not upset 
by the fact that an experiment is going on. 

Even with these precautions, however, it is not clear whether the 
pulse rate could provide us with differential rates for different 
methods of doing a task or different conditions of work. Most of the 
experimentation has been concerned with the question of whether or 
not the pulse is affected significantly by any sort of mental work, in 
comparison with rest or no work. 

One study of visual fatigue (a special topic to be discussed more 
fully later) throws some light upon this question of the differential 
effect of difficulty of the task. Bitterman ^ found that pulse rate was 
reliably increased by a number checking test even with a relatively 
short rest period (fifteen to twenty minutes) before the work began. 
In reading a popular book, however, the pulse rate was decreased 
during reading, and fell throughout the reading period. This might 
be interpreted as a difference of effort, since one task involved a speed 
requirement while the other involved reading an interesting book at 
a comfortable pace. 

Other results, however, failed to show that pulse rate is differen- 
tially significant. The experiment on number checking consisted of 
two parts, in one of which the test was in smaller type than in the 
other. Both tests increased the pulse rate significantly over resting 
values, but the difference in effect was not significant. The average 
pulse rates were, respectively : rest, 73.3 ; large type, 79.3 ; small type, 
80.0 This was in spite of the fact that the small type was sufficiently 
more difficult to change the rate of work (test scores) in a significant 

It is very likely that a useful circulatory index will involve both 
heart rate and blood pressure, as Lovekin has proposed.® 

Lovekin used the combined index of ptdse-prodn-ct (pulse rate 
multiplied by the pulse pressure) as an index of effort. In experi- 
ments with small numbers of subjects, this index was correlated with 
oxygen consumption and lactic acid formation in muscular work. 
The relationship was considered close. In very brief experiments 
with other activities, the pulse pressure appeared to be correlated 
with emotion and attention. 

^ M. E. Bitterman, Heart rate and frequency of blinking- as indices of ^'^sual 
efficiency, Journal of Experimental Psychology, 1945, 35, 279-292. 

Pulse rate is affected significantly by noise, however. Cf. M. E. Bitterman 
and E. Soloway, The relation between frequency of blinking and effort expended 
in mental work. Journal of Experimental Psychology, 1946, 36, 134-136. 

8 O. S. Lovekin, The quantitative measurement of human efficiency under factory 
conditions, Journal of Industrial Hygkm md Toxicology, 1930, 12, 99-120, 153-167, 

Ch. 5] 


Lovekin applied the pulse-product index to workers in factory 
production. After allowing for the effects of meals and rests, he de- 
cided that the general trend in pulse-product was downward through- 
out the working day. Fatigue was indicated by decreasing production 
and decreasing pulse-product. An index of efficiency obtained by di- 
viding the production rate by the pulse-product at each period of the 
day indicated no consistent trends in efficiency as the working day 
progressed. In fact, the highest efficiency occurred in the ninth, out 
of the ten periods into which he divided the day. 

The same author gives an example of the use of this efficiency 
index for comparing different methods of work. In a certain task, 
total production was higher when there were no rest periods. He 
found, however, that while the production fell off 5 per cent with the 
rest periods, the pulse-product was reduced by 16 per cent. He con- 
cludes, therefore, that the rest periods were more efficient, in spite 
of the reduction in production rates. 

Lovekin's report represents an interesting exploratory study, but 
the evidence which is offered to support the use of pulse-product as 
an index of effort is too scanty to permit us to regard it as anything 
more than an interesting suggestion. 

So far, there is no convenient method for obtaining continuous 
recording of blood pressure. It must be taken periodically and is 
subject to temporary variations dlie to a variety of factors. There- 
fore frequent readings and large numbers of readings are essential 
to significant information upon the circulatory effects of work. The 
readings usually interrupt the work, even if they do not disturb the 

While there is a place for much more research into this general 
problem, the present situation can only lead us to conclude that there 
is no established technique of estimating effort by means of circulatory 

Muscular Tension as an Index of Effort 

There is a strong possibility that measures of muscular contraction 
will furnish us with a sensitive and versatile index of effort. This 
statement can be made in spite of the objections which we have set 
forth in preceding pages against identifying alertness, attention, or 
problem solving with patterns of muscular contraction. We have 
surely not described all there is when we describe the pattern of mus- 
cular activities involved in any given activity, no matter how complete 
this description may be. It is quite likely that the circulatory activi- 



[Ch. 5 

ties are more complex than would be suggested when we think of 
them only as supply lines for the muscles. In addition, little is known 
of the central nervous system's contribution to effort, and even the 
digestive and glandular systems may be involved in some way. 

Nevertheless, muscular tensions constitute one important aspect 
of total bodily functioning; they are to be regarded as an important 
bodily resource in psychological activity. In addition, these tensions 
are more amenable to experimental study than some of the other 
bodily activities. They have been subjected to a great deal of pre- 
liminary research within recent years. ^ 

The following relationships have been studied : 

1. Pressure exerted in performing the movements of a reaction 
test.^^ It is found that noise, for example, tends to increase the pres- 
sure exerted upon the reaction keys. This is taken to mean that 
additional effort is used when the work is carried on under more diffi- 
cult conditions. 

2. Recording of electrical potentials by amplifiers. Electrical po- 
tentials developed in various muscle groups have been recorded di- 
rectly by means of special amplifiers. By means of this technique, 
Davis (as shown in Figure 12) has supported Morgan's conclusion 
concerning increased muscle tension during noise.^^ He has also 
shown that these patterns change with an increase in difficulty of 
the task.^^ Potentials from both arm and neck increase with the 
difficulty of the task. (Those problems which were failed by a larger 
proportion of subjects were considered the more difficult.) 

The pattern of distribution of the tensions in various parts of the 
body varies with the kind of activity in which the subject is engaged.^^ 
During mental arithmetic the greatest amount of muscle activity was 
found in the right arm. The other two regions studied (the other 
arm and one leg) also showed increased activity, but not so much. In 
learning nonsense syllables, the activities of the three regions were 
more nearly equal, the left arm showing the greatest contraction. The 

» See F. A. Courts, Relations between muscular tension and performance, 
Psychological Bulletin, 1942, 39, 347-367, for a summary of this field. See R. C. 
Davis, Methods of measuring muscular tension, ibid., 329-346, for a survey of 
methods of measuring muscular tension. 

10 J. J. B. Morgan, The overcoming of distraction and other resistances, 
Archives of Psychology, 1916, No. 35, p. 84. 

11 R. C. Davis, The relation of certain muscle action potentials to "mental work," 
Indiana University Science Series, No. 5, 1937. 

12 R. C. Davis, The relation of muscle action potentials to difficulty and frus- 
tration. Journal of Experimental Psychology, 1938, 23, 141-158. 

13 R. C. Davis, Patterns of muscular' activity during mental work and their 
constancy, Journal of Experimental Psychology, 1939, 24, 451-465. 

Ch. 5] 



control of the various muscle groups is not, however, independent. 
Davis reports that subjects instructed to relax the right arm during 
mental multiplication or learning relaxed the muscles of other parts 
of the body as well. When the relaxation occurred, the output 
seemed to be reduced. 

While the direct measurement of potentials from various muscle 
groups is the most clearly valid method of measuring muscular tonus 
or tension, it has certain disadvantages. The method is confined to 




\ . / 







1 1 . 

I t 

1 1 1 1 1 

1 1 1 J, 

. 1 

20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 

Figure 12. Action Potentials During Rest and Work (Reading), 

Without Distraction 

(From Davis, op. cit., 1937, p. 24.) 

With and 

laboratory applications because of the amount of apparatus involved, 
and because the subjects must be electrically shielded to prevent po- 
tentials from other sources entering the amplifiers and confusing or 
covering up the records. 

Even in the laboratory, the method is subject to some disadvan- 
tages. The muscle potentials involved in complicated activities may 
be all over the body. The activity of one muscle group cannot be 
taken as representative of what is happening in other muscles. The 
focus of activity may shift from one part of the body to another 
with changes in conditions of work. One muscle group might, there- 
fore, show a high level of activity while the average tension of the 
whole body is low. 

To measure total effort as reflected in muscle tonus it would be 
necessary to record simultaneously the potentials from a number of 
different parts of the body. This would be a difficult and expensive 
process. Perhaps it will be discovered that there are certain muscle 



[Ch. 5 

groups which can be taken as representative of larger areas of the 
body, and thus it will be possible to reduce the number of simultane- 
ous observations required. Much research remains to be done, how- 
ever, before we reach this point in the application of amplifier 
techniques to the study of efficiency. 

Several studies of practical problems by this method have been 
inadequate because they involved only one or two muscles. It has 
been found, for example, that work during loud noise at first raises 
the potentials of the forearm, and that then the potentials gradually 
decrease, but we have no way of telling whether or not the tensions 
have moved to some other part of the body, or discovering that there 
has been a bona fide reduction in over-all tension. This criticism 
would apply equally well to the methods which record pressure exerted 
by the working member which we described previously. 

3. Mechanical recording of muscle contractions. The thickening 
of a muscle can be recorded by means of a lever or other device at- 
tached to the body surface, or the tautness of a tendon measured 
by pressure applied. By these methods results comparable to the 
findings of electrical studies have been reported.^* This method re- 
quires less expensive apparatus, and the subjects do not, of course, 
have to be electrically shielded. The apparatus is likely to restrict the 
movements of the subject and sometimes it has required the subject 
to maintain a somewhat artificial position. In addition, these me- 
chanical methods measure the activities of a single muscle group, 
and so are subject to the same disadvantages as the electrical methods 
described in the preceding paragraphs. 

4. Electrical resistance of the skin. The electrical resistance of 
the skin, as measured by ordinary physical methods, shows variation 
with the activity of the subject. There is now evidence that as the 
subject becomes more active muscularly, as he becomes involved in 
an emotion, or as he engages in mental work, the electrical skin re- 
sistance is reduced. Consequently, skin resistance is possibly an 
index of effort — the lower the resistance, the greater the effort. Let 
us examine the concrete evidence upon which this suggestion is based. 

It has long been known that the electrical resistance of the skin 
falls rapidly after sudden stimuli (loud sounds, electric shocks) or 
when embarrassing questions are asked. In fact, these sudden 
changes in resistance are used as the basis of one of the common 
methods of lie detection. These fluctuations are now generally re- 

1* G. L. Freeman, The spread of neuromuscular activity during mental work. 
Journal of General Psychology, 1931, 5, 479-494. 

Ch. 5] 



garded as a function of sweat gland activity which is controlled by 
the sympathetic nervous system — a portion of the nervous system 
which has much to do also with the control of circulation, digestion, 
and other ''vegetative" functions. The palms of the hands and 
the soles of the feet seem to be more easily influenced than other por- 
tions of the body surface by slight changes in the activity of the 
subject, although the changes do occur in other surfaces also.^® 

As we have indicated, the earlier studies of electrical skin resist- 
ance were concerned with short-term, relatively sudden changes in 
resistance. This phenomenon has become known as the "psychogal- 
vanic response" (PGR). More recently, however, it has been ob- 
served that the general level of resistance varies more gradually, and 
that this general level also has significance. 

The level of skin resistance is now often regarded as dependent 
upon the general level of excitation of the muscles of the body. It is 
therefore hoped that it can provide us with an index of the total 
muscular activity and tension of the body as a supplement to the in- 
formation about specific muscle groups which is given us by other 

This hope is based upon a number of lines of evidence which we 
may examine briefly : 

(a) The resistance changes with gross changes in muscular effort 
or relaxation. For example, the skin resistance is definitely lower when 

15 R. S. Woodworth, Experimental Psychology, New York, Henry Holt & Co., 
1938, pp. 276-297. 

1® C. W. Darrow and G. L. Freeman, Palmar skin resistance changes contrasted 
with non-palmar changes and rate of insensible weight loss. Journal of Experi- 
mental Psychology, 1934, 17, 739-748. In this article the authors conclude that 
palmar resistance changes are correlated with changes in alertness, anticipation, or 
apprehension, while non-palmar changes are correlated with the amount of mus- 
cular activity. That is, the non-palmar sweat glands are mechanisms of heat dis- 
sipation, while the palmar glands have a different function. In a later article, 
however. Freeman and R. M. Simpson (The effect of experimentally induced mus- 
cular tension upon palmar skin resistance. Journal of General Psychology, 1938, 18, 
319-326) apparently contradict this conclusion, stating that "the different skin areas 
represent largely differences in the ease of excitation, with the palm becoming more 
active first and dropping toward its limit of physiological reactivity as the non-palmar 
areas become increasingly responsive." This is in spite of the definite statement 
of Darrow and Freeman in the previous article that they did not find that the 
palmar resistance was always low when the non-palmar areas became active. 

This question has an important bearing upon the question of what the skin re- 
sistance indicates, and the role of muscle tension in psychological activity. If we 
identify psychological activity and muscular activity, then the only question is, how 
best to get an index of activity of the muscles from the skin resistance. If, how- 
ever, we grant that there are other bodily systems involved independently in any 
psychological activity, it might be very important to find out whether the palmar 
resistance indicates a kind of internal activity different from that related to non- 
palmar changes. The palmar changes might, for example, be more closely related 
to circulatory and digestive functions than to skeletal muscular activity. 



[Ch. 5 

the subject is sitting up than it is when he is redining.^^ It is not known 
whether the resistance is proportional to the amount of relaxation or only 
roughly correlated. Wenger and Irwin, for example, found that the 
resistance was not much lower in an extreme state of tension than it was 
when the subject was sitting up. Darrow finds that a relatively slight 
change at low resistance levels may indicate as much physiological change 
as a large variation when the resistance is high.^® He therefore pro- 
posed the use of the logarithm of the conductance (1/R), believing that 
this measure is more nearly proportional to the tension than resistance 
(R) as measured in ohms.. 

It may be that it will be necessary to use the palmar resistance for 
some part of the total range of tension, while using non-palmar resist- 
ance to indicate changes in the more extreme parts of the range.^^ 

(b) The skin resistance falls during sedentary activities of various 
kinds (mental arithmetic, solving problems from an intelligence test, 
etc.). There seems to be some evidence that the change in electrical 
resistance is greater if the problem is more difficult, or if the subject is 
"frustrated" in some way. The latter evidence is, however, still incom- 
plete, and much study is needed in order to determine the relationship 
of difficulty of the task, or the effort required, and the level of resistance. 
Freeman, for example, compared the resistances while the subject 
worked at a "comfortable pace" which the subject had chosen, with the 
resistances (luring work at a slower or faster pace.^° While all the 
differences were in the expected direction, none of them were fully 
significant statistically. In another experiment involving sensory dis- 
criminations, it was possible to measure the difficulty of the discrimina- 
tions in terms of the number of failures in a certain comparison. Here 
there was a clear-cut correlation between level of difficulty and lowered 
resistance, but no statistical analysis is reported. 

Another experiment of Freeman's illustrates another phase of this 
relationship. 2 2 In an extended series of reaction time measurements 

M. A. Wenger and O. C. Irwin, Fluctuations in skin resistance of infants 
and adults, and their relation to muscular processes, Iowa University Studies in 
Child Welfare, 1936, 12, No. 1, 143-179. 

18 C. W. Darrow, The equation of the galvanic skin reflex: I. The dynamics 
of reaction in relation to excitation background, Journal of General Psychology, 
1937, 16, 285-309. 

19 See footnote 16, p. 107. 

2 G. L. Freeman, The relationship between performance level and bodily 
activity level, Journal of Experimental Psychology, 1940, 26, 602-608. 

21 G. L. Freeman and W. J. Giese, The relationship between task difficulty and 
palmar skin resistance, Journal of General Psychology, 1940, 23, 217-220. The 
result here may be a special one, since the discriminations were given in ascending 
order of difficulty within the same period. 

22 G. L. Freeman, op. cit., 1940. 

Ch. 5] 


with the same subject, it was found that the fastest average reaction 
time accompanied a moderately low resistance. If the resistance dropped 
still lower, the reactions were irregular and the average time was slower. 
The reactions were also slow at those times when the resistance was high. 
Thus we cannot determine the relationship of resistance to effort by 
measuring output. The individual may be working hard, perhaps too 
hard, and still show a poor performance. The difficulty of the task, or 
the effort required, must be measured in some manner independent of 
the output of any given subject. 

In addition to these lines of evidence upon specific methods of 
measuring tension, there have been a number of researches concerned 
with the effect of tension upon performance. It has been found, for 
example, that rate of learning and performance on some other tasks 
can be influenced by artificially adding to the total muscular tension of 
the subject. If, while working, the subject is. asked to compress a 
spring or support a weight with one limb, his performance may be 
increased unless the additional tension becomes too great. 

Thus it appears that even extraneous tension which might appear 
to be "waste effort" may enhance performance. Just what the mech- 
anisms are is a subject of considerable discussion which need not 
concern us here. 

Bills takes the position that efficiency depends upon the general 
tension level of the body.^* If this level is either too low or too high, 
efficiency goes down. It is certainly indicated by these experiments 
that efficiency can be impaired by levels of tension which are too high. 
Here, more effort is being expended while the output is reduced. 
When output is reduced by too low a level of tension, the net influence 
upon efficiency is less clear. The net effect depends upon the relative 
loss in output and the amount of reduction in tension. Thus we may 
say that a person under induced tension learns faster than he does 
under normal tension, but to say that he is more efficient goes beyond 
the information available. There is a good deal of confusion in dis- 
cussions of learning because in the context of learning studies time 
required to learn has often been taken as the sole element of the cost 
of work. This is, of course, merely equating efficiency and rate of 

2 3 F. A. Courts, Relations between muscular tension and performance, Psycho- 
logical Bulletin, 1942, 39, 347-367. 

24 Bills, op. cit., pp. 22-25. The conclusion that too little tension also decreases 
efficiency may be a confusion of the terms efficiency and output in Bills' discussion 
of this point. (Bills does not include "efficiency" in his glossary of terms or in 
his index, and seems to vary somewhat in his use of the term.) 



[Ch. 5 

Summary. — From the evidence now available, it appears that 
measurements of muscular tension furnish us with our most prom- 
ising line of approach to the measurement of effort in those tasks which 
are commonly classified as light manual work and mental work. 
Electrical potentials from the muscles can be measured in the labora- 
tory, where special foci of muscle contraction are of interest. It may 
be that some form of composite measurement will permit us to evalu- 
ate the total muscular activity during work, just as we can get a 
measure of the average temperature of the whole surface of the body. 

There is also considerable promise in electrical skin resistance as 
an index of the over-all activity level, with the resistance going down 
as the activity increases. Whether it indicates muscular tensions 
alone, or whether it reflects the general level of internal activity of 
all kinds, it is probably a useful index, since all these activities are to 
be reckoned in the cost of work. In fact the index might be even 
more useful if it reflects more than the level of muscle tension alone. 

Skin resistance can probably be used immediately if it is coupled 
with other evidence as a check. There is still much fundamental re- 
search to be done in determining exactly what it measures, how much 
influence relatively unimportant factors may have upon the level of 
resistance, and how accurate it is as a measure of amount of effort 
as well as a reflection of relative changes of effort. 

Some Additional Physiological Changes — Heavy muscular work 
produces many changes in physiological functions besides those which 
we have touched upon in this book. It is natural that some of these 
changes should be studied in connection with sedentary work as well. 
As an example, we may mention briefly the studies of blood count as 
affected by work, since some significant effects of sedentary work 
have been claimed in this aspect of function. 

Muscular work produces increases in the numbers of both red and 
white blood cells, with some evidence of a relative as well as an ab- 
solute increase in the number of white cells (leucocytosis).^^ Moss 
and his collaborators found that fifteen minutes of activity on the 
bicycle ergometer increased the white cell count by more than 50 per 
cent.^^ It has been claimed that ''mental" work also produces leuco- 
cytosis, perhaps even more marked than that observed in muscular 
work.^^ Laboratory studies involving psychological test activity for 

25 Atzler, op. cit., 1938, 372-375. 

2 6 F. A. Moss, J. H. Roe, O. B. Hunter, L. French, and T. Hunt, The meas- 
urement of fatigue by physiological methods, Journal of Experimental Psychology, 
1931, 14, 423-438. 

27 R. M. Skliankaia, C. E. Shakhnovich, A. E. Levina, and A. I. Zak, Changes 
in the composition of blood during mental and light physical work. Journal of 

Ch. 5] 



two hours, as well as occupational studies of telephone operators and 
hosiery knitters, were the basis of these conclusions. Even in these 
positive results, however, there are irregularities which signal caution 
in interpretation. The hosiery knitters, for example, failed to show 
the increase in white cells on the night shift, and the authors point 
out the great irregularity of changes from individual to individuals^ 
There are, perhaps, diurnal variations as well as other unpredictable 
factors to be taken into account. In addition, there is little correlation 
with the duration or amount of work performed. 

In a study of truck drivers who had been on the road for varying 
numbers of hours, the white count for those who had been driving 
was compared with the count for those who had not been driving. 
While there was a greater number of white cells for those who had 
been driving as compared with those who had not driven at all, there 
was no clear change in the white count as the number of hours of 
driving increased. In some groups, in fact, the group who had been 
driving for ten hours showed a lower count than those who had 
been driving for shorter periods. 

Besides these irregularities of the effect of sedentary work upon 
the blood count, irregularities which make the count a doubtful 
measure of effort or of fatigue, there have also been researches with 
completely negative findings. Dunajewski and Kaplan could find no 
effect of three hours of work at various sedentary tasks upon ten 
subjects. At least there was no effect which was greater than the 
normal "random" variations in the blood structure. 

Since changes in internal bodily economy are also reflected in the 
chemical composition of the urine, a number of tests involving urine- 
analysis have been employed. Until recently, however, there appeared 
to be insufficient regularity in these findings to provide a quantitative 
indicator of the cost of work, especially in the more sedentary types 
of occupation. Pincus and Hoagland have reported promising results 
with a new type of test, based upon analysis of certain urinary com- 

Industrial Hygiene and Toxicology, 1938, 20, 169-178. G. A. Lewitina, A. J. 
Lewina, O. S. Tschernomordik, K. S. Samytschkina, L. M. Sidorowa, and S. S. 
Shapiro, Der Einfluss geistiger Arbeit auf das weisse Blutbild. Ein Beitrag zur 
Frage des neurogenen Ursprungs der Verschiebung in der Schillingschen Formel, 
Arbeitsphysiologie, 1931-1932, 5, 115-124. 

28 Skliankaia, et al., op cit. 

29 B. F. Jones, R. H. Flinn, and E. C. Hammond, Fatigue and hours of service 
of interstate truck drivers, U. S. Public Health Service, Public Health Bulletin 
No. 265, 1941, 53-57, 227-240. 

3 M. J. Dunajewski and P. M. Kaplan, Uber das Blutbild bei geistiger Arbeit, 
Arbeitsphysiologie, 1932-1933, 6, 437-444. 



[Ch. 5 

ponents derived from glandular secretions.^^ These products derive 
from the adrenal cortex and are called 17-ketosteroids. Products of 
this gland have long been known to relate to the processes of muscular 
activity, fatigue, and metabolism. 

The amount of steroid excreted in each sample is compared with 
the amount excreted at the same time of day on a control or non- 
working day, since the excretion varies from a high rate in the morn- 
ing to a low rate at night. In studying the effect of work, this normal 
variation is allowed for, and the excess is correlated with the activities 
of the experimental subject.^^ The experimenters found that a pro- 
longed activity upon a test which simulated the activities of flying 
produced a marked excess of ketosteroid excretion toward the end of 
the period. This excess was correlated with a parallel increase in 
errors in the test. Later, similar results were observed in pilot in- 
structors and test pilots actually engaged in flight. In the flying 
instructors, the correlation between excess ketosteroid and per cent of 
the time spent in the air was high (r = .978). The individual pilot's 
rated ability to ''take it" was also correlated with his characteristic 
individual rate of ketosteroid excretion (r=.676). It was also 
noted that the total urinary output was correlated with the amount of 
flying time or time at work in the flying test. The relationship of 
this diuresis to errors or to flying time was not so close, however, as 
that observed for the ketosteroid excess. As a further check upon 
the validity of their findings, Pincus and Hoagland have performed 
controlled studies of the effect of administering a synthetic steroid to 
working subjects. By this means they were able to reduce the 
errors in the flying test which normally occurred toward the end to the 
test period. They have also found significant increases in factory 
productivity resulting from the administration of this substance. 

It would appear that this discovery will be valuable for the study 
of occupations involving strong emotional stress coupled with exact- 
ing, controlled muscular activity. That is as far as we are justified in 
going at the present time. To show that the test is useful for indus- 

31 G. Pincus and H. Hoagland, Steroid excretion and the stress of flying, 
Journal of Aviation Medicine, 1943, 14, 173-193. H. Hoagland, Adventures in 
biological engineering, Science, 1944, 100 No. 2587, 63-67. 

3 2 The "normal" curve is based on results for seven subjects in ordinary daily 
activities (uncontrolled) ; Pincus and Hoagland, ibid. 

3 3 G. Pincus and H. Hoagland, Effects of administered pregnenolone on fatiguing 
psychomotor performance, Journal of Aviation Medicine, 1944, 15, 98-115, 135. 

3* G. Pincus and H. Hoagland, Effects on industrial production of the adminis- 
tration of A 5 pregnenolone to factory workers: I, Psychosomatic Medicine, 1945, 
7, 342-346. G. Pincus, H. Hoagland, C. H. Wilson, and N. J. Fay, Effects on 
industrial production of the administration of A 5 pregnenolone to factory workers : 
II, ibid., 347-352. 

Ch. 5] 



trial tasks, we should need answers to the following questions : 
(1) Do similar excretions occur as regularly in types of work in 
which the stress is due to distaste for the work, awkward postures, 
and static strains, ''intellectual" work or problem solving, and so on. 
The results of administering steroids do not indicate the sensitivity of 
steroid output as an index of fatigue. (2) Are the individual differ- 
ences in excretion so great that we must use very large groups of 
subjects in a study of methods of work, or the effects of certain fac- 
tors upon work? (3) How representative is the ''normal" curve 
which these investigators use in computing all excess values ? Varia- 
tions in this "normal" curve might produce a spurious appearance of 
excess excretion toward the end of the day. This could not account 
for the results when excretion was correlated with percentage of time 
spent in flying, unless there were a tendency for more flight in the 
later hours. In other studies, however, excess is correlated with time 
spent in work. If there is at the same time a progressive error in 
computing excesses, it would produce a spurious correlation. (4) Is 
the indicator sensitive enough to reflect variations in a single factor, 
or will it reflect only changes in several conditions at once? Pincus 
and Hoagland have shown that the steroid excess is increased when 
the work is carried out with a serious deficiency of oxygen. Would it 
be similarly affected by changes in temperature, lighting, noise, and 
the like, and how sensitive would it be to variations in these factors ? 
(5) How cumbersome and expensive is the method? 

These authors have mentioned that they were making some indus- 
trial studies which may help to answer some of these questions. At 
the present writing these results have not yet become available. 

Another measure, which falls somewhere on the border line be- 
tween performance tests and physiological measures, is the critical 
flicker-fusion frequency. This is measured by asking the subject to 
watch an area which is illuminated intermittently. The frequency of 
the flashes of light is gradually increased until the subject reports that 
he no longer sees any flicker. The test may also be conducted with 
varying brightness of the flickering light as well as with varying 
frequency, since the perception of flicker is a function of both vari- 
ables. As the brightness increases the critical frequency necessary to 
eliminate flicker also increases. (See Figure 13.) 

35 This fifth question is not, of course, of equal importance with the other four. 
If a highly reliable and valid technique can be discovered, the gains from its use 
would ultimately justify its expense if no other method were available. Neverthe- 
less, the speed with which advances can be made will depend very definitely upon 
this factor. 



[Ch. 5 

The test originated as a device for studying certain aspects of the 
visual mechanism, but it has recently been brought into the context 
of fatigue. Interested in the possible fatiguing effects of motion pic- 
tures upon the ocular mechanism, Snell measured the critical fre- 
quency before and after a period of exposure to a flickering 

I 1 1 1 1 1 \ ■ — i I 


Figure 13. The Relation between Critical Flicker-Fusion Frequency and Brightness 

Lower and upper curves for areas of 6" and 19°, respectively, from data of Hecht and 
Smith. Middle curve for area of 12° for one of Lee's subjects. 

(From R. H. Lee, Critical fusion frequency of flicker, U. S. Public Health Service, 
Public Health Bulletin No. 265, 1941, p. 195.) 

Stimulus. He found that exposure to visible flicker had the effect of 
reducing the critical frequency; that is, a rate of alternation which 
would be seen as flickering before the exposure would later appear to 
be continuous. According to Snell, it made no difference what rate 
of flicker produced the ''fatigue," so long as it was visibly flickering. 
This finding is supported in general by McFarland and his collabor- 
ators, although they varied the intensity of the test stimulus rather 

3 6 p. A. Snell, An introduction to the study of visual fatigue, Journal of the 
Society of Motion Picture Engineers, 1933, 20, 367-390. 

Ch. 5] 



than its frequency. The latter investigators found that a slowly 
flickering stimulus produced a greater effect in a given time interval 
than a more rapidly flickering light. 

More recently, Simonson and Enzer have proposed the use of the 
flicker fusion test as a measure of "general" fatigue of the central 
nervous system, removing the test from the context of specific ocular 
fatigue. In a number of office, laboratory, and professional workers 
they found a decrease of the flicker fusion frequency in the afternoon 
as compared with the morning value, the decrease occurring in every 
case they studied. They also found that benzedrine administered to 
the subjects during the day always eliminated the effect of the day's 
work upon the fusion frequency. 

Simonson and Enzer's findings were so uniform that they gave 
great promise to the technique of flicker fusion testing. Unfortu- 
nately, however, their results were not duplicated in a similar experi- 
ment carried out by McFarland and his collaborators.^*^ Seven of the 
nine subjects used by McFarland were engaged in reading examina- 
tions for the whole working day, and the other two were employed 
in computing and drawing. In most cases there was no difference in 
the results of the flicker test in the afternoon as compared with the 
morning. Those differences which did occur were in opposite direc- 
tions with about equal frequency. 

There was only one result in McFarland's study which would 
support Simonson's contention that the flicker fusion test is a measure 
of fatigue of the central nervous system (or of any sort of fatigue in 
sedentary work). Subjects who had gone without sleep for a period 
of at least twenty-five hours showed some individually significant 
changes in the frequency-intensity relationships of the flicker test. 
(The five tests upon three subjects were divided as follows : One sub- 
ject, tested once, showed no change following the sleepless period. 
A second subject, tested twice, showed a significant change on one 
test but not on the other. The third subject showed significant 
changes on both occasions.) On the basis of these rather meager 
results, McFarland concludes that "loss of sleep appears to produce a 
significant change in the flicker curves. Following loss of sleep, a 

37 R. A. McFarland, A. H. Holway, and L. M, Hurvich, Studies in Visual 
Fatigue, Harvard University: Graduate School of Business Administration, 1942, 
pp. 191-219. 

3 8 E, Simonson and N. Enzer, Measurement of fusion frequency of flicker as a 
test of fatigue of the central nervous system. Journal of Industrial Hygiene and 
Toxicology, 1941, 23, 83-89. 

39 E. Simonson, N. Enzer, and S. S. Blankenstein, Effect of amphetamine (ben- 
zedrine) on fatigue of the central nervous system. War Medicine, 1941, 1, 690-695. 

4 R. A. McFarland, et al., op. cit., 196-204. 



[Ch. 5 

greater intensity is necessary at a fixed frequency for flicker recogni- 
tion to occur throughout most of the range used. There are, how- 
ever, considerable individual differences." 

It is possible that some of the differences in results of Simonson 
and of McFarland are attributable to differences in technique. Simon- 
son varied the frequency of the flashes at certain fixed intensities of 
the light, and determined the threshold of recognition of flicker in 
terms of frequency. McFarland varied the intensity at certain fixed 
frequencies to determine thresholds in terms of intensity. A curve 
like that shown in Figure 13 could, of course, be plotted from either 
kind of data, since it would not matter which variable is considered 
dependent. The difference in technique may, however, make a real 
difference to the subjects, and it is therefore still possible that fatigue 
would show its effects upon one type of threshold and not upon the 
other.^^ It is unfortunate, therefore, that the tests have not been 
carried out by both methods upon the same subjects and under the 
same working conditions. 

Using a method more closely similar to that of Simonson, Lee 
measured the fusion frequencies of truck drivers who had been driv- 
ing for varying periods before the test.*^ (This study was a part of 
the program mentioned previously in connection with blood tests for 
fatigue (page 110) ; it is discussed again in a later chapter on acci- 
dent problems (Chapter 12). There was a slight decrease in the 
critical frequency as the number of hours of driving increased, but 
none of the results were statistically significant. Again it is difficult 
to interpret the results because there was no control test before 
driving for the specific driver. One group of drivers had not driven 
at all, while another group had driven before the test. The failure to 
find significant differences may, therefore, be due to the increased 
variability of the data resulting from individual differences. It is 
notable that the differences between drivers in different cities were 
almost as large and as consistent as the differences between drivers 
grouped according to the number of hours of previous work. This 
suggests other important variables which have not yet been taken 
into account, and which must be discovered before the results of such 
tests would be meaningful. 

41 O/'. cit., p. 217. 

*2 That is, even if the two methods give equivalent results for unfatigued sub- 
jects, fatigue might have different effects upon the two kinds of judgment. 

*3 R. H. Lee, Critical fusion frequency of flicker, in B. F. Jones, R. H. Flinn, 
et al.. Fatigue and hours of service of interstate truck drivers, U. S. Public Health 
Service, Public Health Bulletin No. 265, 195-208. 

Ch. 5] 



An entirely different application of the flicker-fusion test was tried 
out by Brozek and Keys.** Since Simonson had supported the theory 
that flicker is affected by general fatigue of the central nervous sys- 
tem, Brozek and Keys measured fusion frequencies for subjects who 
were active in heavy muscular tasks. They first measured the relia- 
bility of the test by comparing repeated measurements for the same 
subjects, and found the test to be highly consistent and unaffected by 
practice. Heavy muscular work, however, had no effect upon the 
fusion frequency except under very extreme conditions. The fre- 
quency decreased slightly from day to day when the subject was 
engaged in work nearly at capacity in a temperature of 120° F. Even 
then, however, there was no significant change from morning to night 
of the same day. There was a similar tendency for a long-term de- 
crease for subjects who were working hard with inadequate food 
intake. None of the changes observed in these studies were large, and 
frequently they were not statistically significant. 

Another study indicates that flicker fusion is related in some fash- 
ion to general bodily economy. In analyzing the effects of oxygen 
deficiency in altitude chambers which simulate the conditions of flight 
at varying altitudes, it was found that flicker fusion frequency was 
reduced.*^ The changes were significant at an ''altitude" of 10,000 
feet. For higher altitudes, there was also a significant residual effect, 
in that the frequency remained low for a time after returning to ''sea 

We may conclude this brief summary of the flicker-fusion test by 
stating that, in spite of the conflicting results now available, it is a 
subject worthy of further serious and well-controlled research. Per- 
haps the test will be useful only in studies of certain visual tasks. 
Perhaps Simonson's belief in its value as a measure of fatigue of the 
central nervous system will be further supported by other kinds of 
experiments. At any rate, the test will be worth the effort which may 
be devoted to it, since it possesses important advantages over many 
performance tests. The subject has no way of judging the accuracy 
of his own performance — in fact there are no "correct" or "incor- 
rect" reactions, and even if there were the subject cannot guess the 
way in which the apparatus is manipulated by the experimenter. It 
is likely, also, that the motivation of the subject would have less effect 

J. Brozek and A. Keys, Flicker fusion frequency as a test of fatigue, Journal 
of Industrial Hygiene and Toxicology, 1944, 26, 169-174. 

*5 J. E. Birren, M. B. Fisher, E. Vollmer, and B. G. King, Effects of anoxia on 
performance at several simulated altitudes, Journal of Experimental Psychology, 
1946, 26, 35-49. 


upon these results than it would for most performance tests. It is 
primarily for this reason that we have discussed the test in the context 
of physiological measures. 

Gross Statistical Measures of Efficiency 

The measures of cost of work and of efficiency which we have con- 
sidered so far are expensive and difficult to apply. They look toward 
future solutions of practical problems by analyzing the effect of work 
upon bodily processes of all kinds. Where more immediate answers 
to practical questions of efficiency have been desired, students of in- 
dustrial problems have often resorted to cruder measures which are 
more easily applied. Thus the effects of rest periods, various lengths 
of the working day, changes in lighting or ventilation methods, and 
of supervisory practices have been studied more frequently w^ithout 
utilizing any of the methods so far discussed, except, possibly, for an 
analysis of daily work-curve patterns. 

Several of the measures listed below would appear to confuse effi- 
ciency with rate of performance. In fact, that confusion has often 
been made in industrial studies. In other cases, how^ever, the interest 
has centered in, efficiency in the more correct sense of the term, and 
these measures can be justified as inexact indices of efficiency under 
certain conditions. There are so many possible measures that we 
shall merely list them briefly, and then consider the arguments which 
may be offered to justify the more important ones : 

1. Long-term trends in production rate. 

2. Records of errors, waste, scrap, etc. 

3. Accident rates. 

4. Turnover rates. 

5. Records of health and absenteeism. 

For all these indices, a prime requisite is adequate statistical treat- 
ment beyond the mere collection of numbers and averages. This 
seems self-evident, but in practice the requirement is often overlooked. 
There are instances on record in which the optimal arrangement of 
rest periods was determined from a small group of subjects and with- 
out any analysis of the variability of these subjects or their repre- 
sentativeness. In other words, it would not be possible to conclude 
that the results are significant for that group, to say nothing of the 
application of the results to other individuals doing the same kind 
of work. 

Ch. 5] 



The reader may feel that we harp upon this theme to an excessive 
degree. It is apparently necessary, however, to emphasize that statis- 
tical treatment to determine the significance of results in relation to 
chance is always essential. 

With this general condition, let us examine some of these indices 
in greater detail in order to note the assumptions upon which their 
use is based, and to evaluate these assumptions. 

L Long-Term Trends in Production Rate. Assuming that there 
has been no change in motivation, it is argued that production must be 
increased as a result of lowered cost per unit. If factors other than 
learning are under study, it would also be necessary to ensure that the 
influence of practice has been eliminated. Ordinarily, this is not taken 
into account, apparently because it is believed that the workers have 
already reached the limits of improvement due to practice. 

The question here is, what is meant by "constant motivation." Usually 
it is assumed that it is the same if the pay remains the same, and if there 
is no evidence of marked changes in supervisory methods, etc. It is 
assumed that under these conditions the worker will expend a fixed 
amount of energy per unit of time. If the rate of production increases, 
it must mean an increase in efficiency. 

There are two possible flaws in this reasoning. First, the factor under 
study may itself provide a change in motivation so that the men are 
working harder after this factor has been introduced. There is no way 
of determining whether or not this is the case without a more direct 
measure of effort expended. Secondly it might be equally reasonable to 
assume that constant motivation implies that the worker would aim at a 
given amount of production or a given amount of money in the pay 
envelope. In other words, the factor under study may have more effect 
upon effort expended than it does upon the gross output. If so, the 
factor would appear to have no effect if our only measure is the output 

There are two general arguments which might be used as a retort 
when we mention the above objections to studying efBciency by analyzing 
production trends. 

(a) The attitude of industrial managers may be that they are not 
interested in these problems. If production goes up, they do not care 
whether it is due to a motivational effect or to an increase in efficiency. 
They would also argue that an increase in effort cannot be seriously 
harmful if the increase in output is maintained over a long period of 
time. A temporary increase followed by a slump would, however, indi- 



[Ch. 5 

cate that the method or factor involved was so inefficient as to be value- 

(b) If, on the other hand, production remains the same, management 
might not feel that it could justify any expense involved in using the 
new factor or method, even if the worker's effort was reduced. 

The first of the above points is probably justified, so long as the pro- 
duction index is followed over a long period of months or years. This 
would mean, however, that the determination of the effects of a given 
factor would take an extremely long time. The second point, that man- 
agement is not interested in reducing the effort unless it is reflected in 
output is not so easily justified. If the effort required for a unit of work 
can be reduced, it should be to management's advantage to do so. For 
one thing, it would probably pay in terms of reduced overhead or indirect 
costs such as turnover, complaints, discontent, and the like. Another 
reason why it might pay is that it provides an additional reservoir of 
productive capacity which could be tapped if suitable incentives were 
provided along with the changed conditions of work. 

We may conclude, therefore, that if a given factor increases produc- 
tion consistently over a long period of time, efficiency has probably been 
increased ; or, at least, that management would be justified in using that 
factor. If, on the other hand, no increase is observed, no clear conclu- 
sion is possible. Production trends, then, are useful as positive evidence, 
but not as negative evidence unless other indices are also studied. 

2. Records of Errors, Waste, Scrap, Etc. If production rates are 
controlled by automatic machinery, or are otherwise insensitive to 
changed conditions, the effects of the changes may still be evidenced in 
measures of the quality of work. We have already indicated that such 
measures of accuracy and control are valuable in laboratory studies 
of production rates. Even if production rates are significantly affected, 
records of quality should be examined to make certain that the changed 
performance is a bona fide change in useful output. 

3. Accident Rates. Although this is usually considered as a sep- 
arate problem, accidents nevertheless represent an evidence of ineffi- 
ciency. In occupations involving frequent or serious accidents, they may 
represent the largest part of the total cost of the work, and energy cost 
would be a relatively minor factor. 

In certain occupations, then, accident rate can provide a useful indi- 
cation of efficiency. There can be no doubt of its validity if records are 
carefully and accurately kept. A long-term collection of data is neces- 
sary, however, in order to ensure representative data. In many occupa- 
tions, the accident rate would be so low that it would be almost impossible 

Ch. 5] 



for practical reasons to obtain sufficient data to measure the effect of 
any one factor. The longer the span of time involved in a set of data, 
the more chance there is that other variables have changed during the 
period of study, and the more difficult becomes the evaluation of the 
findings for this reason. Accident rate can be considered only as a spe- 
cial index to be applied to a limited number of occupations. 

4. Turnover Rates. Here again we meet the difficulty that observa- 
tions must be carried on over a long span of time if they are to be sig- 
nificant. In addition, turnover rates are subject to even more variations, 
due to extraneous factors which are not the immediate subject of investi- 
gation. Is it often difficult to separate motivational factors from fatigue 
and other elements involving capacity for work. 

From the point of view of management, turnover rates evidently re- 
flect the over-all efficiency of the organization's supervision, recruitment, 
or other personnel policies. Our more immediate concern is with the 
efficiency of performance of the worker himself. It is probable that an 
increase in turnover rate would constitute a gross indication that there is 
some increase in discontent, fatigue, or energy expenditure. It would 
indicate decreased efficiency, although without other information it would 
be impossible to determine what elements of the cost of work are pri- 
marily responsible for the change. 

5. Records of Health and Absenteeism. The incidence of illness 
or of some particular disease is even less sensitive to variations in some 
factor in the work than other measures mentioned above. Of course, 
specific immunological treatment should show its effect upon large 
groups in a reasonable period of time. But if it is desired to find the 
effect of such a factor as overtime or long working days upon the inci- 
dence of illness, a much more extensive sample would have to be studied, 
and over a much longer period of time. If absenteeism is to be consid- 
ered as a reflection of effects upon health, then some form of investiga- 
tion of each case to determine the cause of absence would be necessary 
to make the data meaningful. 

Each of these gross measures of efficiency is useful, therefore, 
under certain conditions, but only limited conclusions can be drawn. 
To get the maximum use from them we hope that they can eventually 
be studied in parallel with more exact analyses of the effect of the 
work upon the worker. The gross measures will continue to be useful 
on this basis, even after more accurate methods are available, because 
they will furnish us with an economical method of preliminary 
analysis of the problem at hand. 



[Ch. 5 

Summary and Conclusions on Measuring the Cost of Work 

The reader has probably gained the impression that these chapters 
on the fundamental problems of efficiency have raised more questions 
than they have answered. If so, it is all to the good. One cannot 
have an adequate grasp of this field until he is aware of the problems 
which need to be solved in paving the way for more adequate solutions 
of practical problems. 

Applied psychology is concerned primarily with practical problems, 
but its aim is to solve these problems by the methods and results of 
scientific investigation. We cannot give satisfactory answers to 
many practical problems of efficiency until the methods of finding 
these answers are developed and established. The methods which 
have been critically surveyed in these chapters provide a beginning 
from which tentative answers to some of these questions can be 

In succeeding chapters we shall consider specific factors which 
affect productivity and efficiency of human performance. We shall be 
able to deal with many of these very briefly because we have already 
become familiar with the technical problems involved in measuring 
the effect of a given factor. 

Among the measures of the cost of work which we have consid- 
ered there have been three main kinds : ( 1 ) Those which show prom- 
ise of developing into valid indices of fatigue or other phases of the 
cost of work, but which are still in the developmental stage. (2) A 
few measures which have more established positions as measures of 
the cost of work, although their applicability is limited to certain defi- 
nite kinds of work. (3) Crude indices subject to a variety of objec- 
tions, but which are relatively easy to apply, and are used as stopgaps, 
providing us with rough solutions until more refined methods become 
practically useful. 

The first of these classes would include : muscle potentials, skin 
resistance as an indicator of tension, steroid excretion, flicker fusion 
frequency, composite measures of circulatory activity, such as the 
pulse-product, and possibly certain psychological test activities, espe- 
cially those which require delicate co-ordinations, and closely con- 
trolled attention. The second class of more firmly established 
techniques would include, as its principal member, the use of oxygen 
consumption for heavy muscular work, possibly also for moderately 
heavy industrial tasks, but not for sedentary tasks. 

The last class of cruder techniques will, unfortunately, provide us 
with the bulk of the information which is available on several of the 

Ch. 5] 



problems of practical import in industrial work. This class includes 
output decrement curves (for heavy physical labor, or other work 
with maximal effort), errors, variability and accidents during per- 
formance, long-term trends in productivity, and, upon occasion, sta- 
tistics of illness, complaints, or turnover. 

Chapter 6 


In the preceding chapters we have discussed some of the funda- 
mental problems involved in evaluating the efficiency of the organism 
under any given conditions. We shall now survey some of the avail- 
able knowledge concerning factors which affect the rate of work and 
efficiency of performance. In the present chapter we shall discuss con- 
ditions which serve as governing factors in performance, that is, they 
increase or decrease the capacity of the individual to perform, or they 
affect primarily the ease or difficulty of the "work itself rather than 
controlling directly the effort which the individual puts forth in his 

It is true that some of these may, on occasion, play a dual role, 
serving as incentives as well as governors. The primary interest in 
research upon those conditions has been, however, with the governing 
effect. Motivation has been considered a secondary variable to be 
controlled and held constant rather than the major interest of the 
experimenter. Temperature, light, noise, rest periods, or the like 
may themselves influence the ''will to work" or the effort expended 
by the worker. So far as possible, this aspect of their effect is elimi- 
nated from experimental study by special controls. Such control is 
not always complete, as we shall see, but it is important to keep this 
purpose in view as we survey the experimental information on these 


It is common to discuss the effects of noise under the heading of 
''distraction." We have deliberately avoided doing so because the 
experimental literature is limited to meaningless sounds or noise, and 
leaves out of account many other potential sources of distraction. In 
addition, it is safer to speak of the effects of noise than of the effects 
of distraction because the term distraction assumes that the noise has 
an effect which must still be demonstrated. 

The emphasis which is given to noise abatement campaigns indi- 
cates a common assumption that noise is a hindrance to effective work. 




The interest in controlling noise has, however, far outrun the factual 
evidence available. The sources of evidence w^hich we have must be 
divided into two groups because of the distinct methods and condi- 
tions which were used. The first group consists of experiments upon 
productivity and effort of subjects working in a laboratory with high 
motivation and relatively short periods of exertion. The second 
group consists of scattered statistical analyses of production in the 
industrial situation, where it has been possible to introduce some con- 
trol of the noise factor but not of many other variable conditions. 

Laboratory Studies. — If we are interested in the effects of noise 
as a governing factor in performance, we wish to find its influence 
upon subjects who are well-motivated. Effects of noise upon the 
normal working rate of workers in the industrial setting may indicate 
either variations in motivation, or in capacity for performance, or 
both. The laboratory studies are important, therefore, in indicating 
whether or not noise has a necessary effect upon the performance of 
well-motivated workers. 

The brevity of the working period in most of the laboratory 
studies gives little opportunity for the worker to become accustomed 
to the noise. In spite of this, it was soon discovered that the effect 
of intense and irregular sounds was surprisingly small. In many 
cases there would be a short period of reduced performance, but the 
recovery would be very rapid. Occasionally, the subject might actu- 
ally accomplish more under noisy conditions than he did in quiet. 
Thus the influence of noise upon rate of performance in the laboratory 
can be considered as only a small and temporary one. 

It is still possible that long periods of work under especially noisy 
conditions might show more effect upon output rates, due to some 
kind of cumulative influence of the sound. So far there is little ex- 
perimental evidence upon these longer periods. In shorter periods 
the subjects apparently become accustomed to the sounds. 

A number of different lines of evidence suggest, however, that this 
adjustment to the noise is attained at the expense of additional effort 
on the part of the subject. For some reason, the subject tends to 
adopt a certain level of output as his goal, rather than a constant 
level of effort. 

In spite of the evidence that effort is increased during noisy condi- 
tions, there is some indication that the major effect of the noise 
upon effort occurs during the initial minutes of the noisy period, and 
the effort diminishes at the same time that the output is returning 
to normal. If this is true, the adjustment to the noise does not in- 



[Ch. 6 

volve a continued poor efficiency; instead, the total adjustment to the 
noise may be a return to normal efficiency as well as a return to nor- 
mal working rate. The evidence upon this latter point is inconclusive 
as yet, so we shall examine it in order to see what remains to be done 
in solving the problem. 

Let us look briefly at the evidence upon which these generalizations 
are based. The two main lines of evidence are studies of muscular 

450 r 

^ I I I I ! \ I — I \ — I — 


Figure 14. Key-Pressure in Work During Quiet and Noise 

Points represent means of successive groups of 100 reactions each. 
(From data in Morgan, op. cit., p. 44.) 

tension during work, and metabolic cost. Morgan ^ recorded the 
pressure which the subjects exerted upon the reaction keys with which 
they performed their experimental task, and also recorded the breath- 
ing pattern during work. There were changes in both records during 
noise, the pressures increasing and the breathing becoming irregular 
in comparison to the working patterns of the quiet periods. The 
major increase in tension (pressure) occurs, however, at the begin- 
ning of the noise, and there is a definite fall in this tension in a rela- 
tively short time. (See Figure 14.) The change in breathing pattern 
during the period of noise is not clear, but it is suggested that the 

1 J. J. B. Morgan, The overcoming of distraction and other resistances, Archives 
of Psychology, 1916, No. 35, 5. 



irregularities which appear during noise are due to the subject's 
^'talking to himself" in order to avoid distraction. This may increase 
as time goes on. Whether this entails increased cost of the work or 
merely a change in the method of performing it cannot be decided. 

Davis has measured muscle potentials directly in subjects who 
were at rest during quiet and noise, and for brief periods of work 
during noisy conditions.^ Electrical potentials from the forearm in- 
creased when noise began, but soon dropped to their normal values 
again (Figure 15). In the second experiment which involved work 


Figure 15. Average Magnitude of Action Potentials on Successive Days of Stimu- 

(From Davis, op. cit., 1935, p. 13.) 

(difficult reading), the recorded period was only two minutes. Dur- 
ing this time the muscular potentials increased. The period was too 
short, however, to determine the possibilities of adjustment to the 

The effect of noise upon the metabolic cost of work has been studied 
by Laird, for typing ; ^ by Vernon and Warner ; * and by Harmon,^ 
for arithmetical calculation. Of these, the most carefully controlled 

2 R. C. Davis, The muscular tension reflex and two of its modifying conditions, 
Indiana University Publications, 1935, Science Series No. 3, and The relation of 
certain muscular action potentials to "mental work," ibid., 1937, No. 5, 23-24. 

3 D. A. Laird, The measurement of the effects of noise upon working efficiency, 
Journal of Industrial Hygiene, 1927, 9, 431-434. 

* H. M. Vernon and C. G. Warner, Objective and subjective tests for noise. 
Personnel Journal, 1932-1933, 11, 141-149. 

5 F. L. Harmon, The effects of noise upon certain psychological and physio- 
logical processes, Archives of Psychology, 1933, No. 147, 23. 



[Ch. 6 

work was apparently that of Harmon. He found the metaboHc cost 
of adding columns of figures was greater during noise than under 
quiet conditions. It should be remembered, however, that the in- 
creases in metabolism due to the work were not great in either case 
as compared to changes which are brought about by muscular work. 
Since the changes were small, and since there were only two subjects, 
it is doubtful how conclusive these results are, in spite of the many 
experimental precautions used by Harmon. 

Several of these experiments have also included records of a meas- 
ure of cost taken upon successive periods of exposure to the noise. 
Davis' experiments show a clear-cut decrease in the effect of noise 
upon muscle tensions, from day to day of his experiment. Harmon's 
subjects showed smaller and smaller effects of the noise upon oxygen 
consumption, as the observations were repeated several times. 

Thus all the facts that are available suggest that there is little 
effect of noise either upon output or upon effort expenditure during 
work after the subject has become accustomed to the noise. At least 
this is one obvious interpretation of the known facts. Poffenberger 
has suggested another possible conclusion, namely, that the pattern 
of tensions shifts during the adjustment to the conditions of noise. ^ 
He leaves open the possibility that efficiency is still low, but that the 
change in pattern of effort is such that the effort is no longer so easily 
detected by the measuring devices used in these experiments. 

Poffenberger 's suggestion would be quite plausible if we were con- 
cerned only with Davis' results which involved muscular activity in 
the forearm alone. It is less satisfactory when applied to the other 
results we have described. Morgan's measure of tension involved 
muscles which were actually used in the activity under study, so that 
it seems less likely that tensions in other parts of the body would be 
substituted for those he measured as the process of adjustment to the 
noise went on. Similarly, if the oxygen consumption has any signifi- 
cance at all in these researches, it indicates the over-all muscular 
activity. As Harmon found, this over-all activity diminishes with 
time. The tension pattern might still be influenced by the noise, as 
Poffenberger suggests, but if it is just as economical as the pattern 
in quiet conditions, the pattern would be a minor consideration. 

Such speculation avails little, of course, except to clarify the na- 
ture of the problem which still remains to be solved. We need a study 
of a greater range of factors in the cost of work. A more thorough 

^ A. T. Poffenberger, Principles of Applied Psychology, New York, D. Apple- 
ton-Century Co., Inc., 1942, p. 138. 


study of bodily tension, including potentials from various portions 
of the body, and the skin resistance index of general tension, would 
be important contributions to this topic. We also need a more com- 
plete survey of the variations in these effects with varying durations 
of the noise, both longer experimental sessions and longer series of 
sessions, in order to study more completely the process of adjustment 
to the sounds. 

All that we can conclude from the facts now before us is evidently 
that meaningless, intense noises are likely to reduce output slightly 
when they first begin, and, for that matter, when they first cease. 
During that same time the worker puts forth greater effort and com- 
pensates to a certain extent for the influence of the stimuli. 

Field Studies. — If the rapid adjustment which is found in labora- 
tory studies occurs also in the normal working situation, and if it 
continues in the same direction, we should expect that noise control 
would have little efYect in industry. There is, however, the real pos- 
sibility that such an assumption is unjustified, that the adjustment 
follows a different course, and that other accumulative effects of noise 
appear. Direct study of factory productivity, with reasonable control 
of conditions, is therefore a requirement. Although controls are 
difficult to obtain under practical conditions, nevertheless such studies 
can be made with present techniques. The study of efficiency under 
noisy conditions in the field is, of course, a more difficult problem 
from a methodological as well as practical point of view. 

In spite of the need for field studies of production, and in spite of 
the apparent feasibility of such investigations, there are few signifi- 
cant findings available. In a recent review of this whole topic, Ber- 
rien notes that most of the articles dealing with these studies are 
written for the layman and give little indication of the procedures 
used. Many cannot be evaluated for this reason. He concludes 
further "Only one field study in this country has been completed with 
sufficient control and covering a sufficient time to place confidence in 
the results." ^ Even in this study the statistical significance of the 
findings has not been reported. 

The investigation to which Berrien refers in the above quotation 
(carried out by the Aetna Life Insurance Company) and a study of 
weavers carried out by Weston and Adams in Great Britain ® repre- 

7 F. K. Berrien, The effects of noise, Psychological Bulletin, 1946, 43, 141-161. 

8 H. C. Weston and S. Adams, The effects of noise on the performance of weav- 
ers, Industrial Health Research Board, 1932, Report No. 65, pp. 38-62. 



[Ch. 6 

sent the small store of information which is even partially interpret- 
able. The studies had similar results, and similar criticisms apply to 
the statistical interpretation. 

For details of the field studies we refer to Berrien's excellent re- 
view. We need only say here that the results are consistent in indi- 
cating increased output when noise is reduced, even when the workers 
are accustomed to the noise. It appears that the adjustment to noise is 
never complete. The laboratory investigations therefore need to be 
extended in order to discover the source of the discrepancies. Perhaps 
the short duration of the work period in the laboratory has been con- 
fined to an initial period in which the subject changes his response to 
the noise very rapidly. As a result, the ease of adjustment has been 
exaggerated. There is also the possibility that noise operates pri- 
marily as a motivational factor in the normal situation, a phenomenon 
which does not appear in the laboratory because of the completely 
different goal of the subject. The question is whether a reduction in 
noise, serving only as an added incentive to effort, could continue to 
be effective over a period of months or years, as indicated by the 
field studies. 

Very little is known of the effects of more meaningful sounds, such 
as conversation or radio programs. This lack is due in part to the 
great difficulty an experimenter would find in providing uniform ex- 
perimental conditions. A repetition of the same conversations or 
programs would change their effect, while a change in content of the 
''distracting" material would provide extremely variable conditions 
for the experiment. It is also doubtful whether an artificially ar- 
ranged or ''staged" situation would have any similarity in effect to a 
similar distracting situation as it naturally occurs. At any rate, an 
experiment upon such meaningful sounds would require much more 
data in order to control some of these additional variables by means 
of statistical analysis. Previous researchers probably have also felt 
that it would be better to clear up the relatively simpler problem of 
noise before attempting these more difficult studies. 

Music. — The frequently mentioned benefits of music as an accom- 
paniment of work probably belong more clearly under the topic of 
incentive and interest in the work. The music may provide interest 
in an otherwise routine and dull task, or it may be taken by the 
worker as an evidence that management is interested in providing 
pleasant conditions of work. Wyatt and Langdon investigated the 
effect of music as a part of their larger studies of boredom which will 
be mentioned again in other parts of this book. They concluded that 


the "effect seemed to be directly related to the amount of boredom 
experienced by the different individuals." ^ 

That the effect of the music depends upon the ''set" of the worker 
is also indicated by an experiment of Baker's. He was able to show 
that the music could increase or decrease performance, depending 
upon whether the subject expected the music to be an aid or a hin- 
drance to his work. 

Temperature and Ventilation 

Although the layman is often unaware of it, the problems of tem- 
perature regulation and the problems of ventilation for effective 
working conditions are usually identical. It is only under exceptionl 
conditions that there is any danger of oxygen deficiency or too much 
carbon dioxide due to inadequate ventilation. Many years ago it was 
conclusively demonstrated that the effects of bad air were attributable 
to faulty temperature regulation. In fact, "fresh air" is no more 
beneficial than "stale air" if the fresh air is not properly controlled 
with respect to temperature and the other factors which regulate the 
heat loss from the body. Naturally, we are excepting from this state- 
ment special industrial conditions in which noxious gases, dust, or 
other irritants are a danger. In the absence of these factors, even in 
a tightly closed room containing several people, it would take a long 
time to reduce the oxygen or increase the carbon dioxide to the danger 
point. Long before that, discomforts due to overheating are likely 
to set in. Thus the two topics which are included in this section 
comprise only a single common problem for efficiency in work. 

Because of the interest of the medical profession and physiologists 
in the control of bodily temperature, and the interest of engineers in 
problems related to the design of heating and cooling equipment, a 
great deal of research has been devoted to the bodily changes which 
result from changes in temperature conditions of the environment. 
The American Society of Heating and Ventilating Engineers has 
supported a long series of these experiments. The Pierce Laboratory 
of Hygiene at Yale University has been another productive source of 
information on the physiological and biophysical sides of the problem. 

This is not the place to attempt even a brief summary of the de- 
tailed physiological and physical facts and problems involved. We 

^ S. Wyatt and J. N. Langdon (assisted by F. G. L. Stock), Fatigue and bore- 
dom in repetitive work, Industrial Health Research Board (Great Britain), 1938, 
Report No. 77, p. 73. 

10 K. Baker, Pre-experimental set in distraction experiments, Journal of General 
Psychology, 1937. 16, 471-486. 



[Ch. 6 

shall merely touch briefly upon a few facts which should be kept in 
mind when we turn to the psychological problems involved. 

Let us consider first the physical factors involved. Most of us are 
aware of the importance of humidity and air movement, as well as of 
air temperature, as determinants of comfort and bodily temperature 
regulation. We often fail to realize, however, that another physical 
factor is also important. This is the factor of radiant heat loss or 
gain from our surroundings. The human skin is an efficient radiator 
or absorber of heat energy; in fact, it is nearly what the physicist 
calls a "black body" so far as heat energy is concerned. As a result, 
we can gain or lose heat from surrounding solid objects independently 
of the .temperature of the air itself. 

Because of radiation it is quite possible to remain comfortable in 
indoor clothing with the air temperature only 40° F. or 50° F. so 
long as the heat loss to the air is balanced by a lower rate of absorp- 
tion by other objects. As an example, one engineer constructed a 
small room of polished copper. Because the walls absorbed so little 
of the heat given off by the body through radiation, the air tempera- 
ture had to be quite low if the occupant was to feel comfortable. 
This factor is responsible for the fact that some rooms feel so much 
cooler at (say) 70° F. than do others. A large cold wall or a large 
window surface not only cools the air so that the room is harder to 
maintain at 70° F., but also absorbs radiant heat. 

In order to specify the conditions which are satisfactory for a 
given type of work, therefore, it is necessary to specify four factors ; 
air temperature, humidity, rate of air movement, and radiation (usu- 
ally expressed as the average temperature of the surrounding solid 
objects). Statements that ''ideal" conditions consist of 68° F. and 
50 per cent relative humidity (as we are often told) are inadequate 
because these other conditions are not described, to say nothing of 
the nature of the evidence upon which such recommendations are 

On what basis should standards of temperature and ventilation 
conditions be determined ? Let us consider as briefly as possible the 
possibilities which have so far been investigated. 

L Effects upon physiological processes such as heart rate, blood 
distribution, perspiration, and the like. These measures are better 
suited to finding the extreme range of temperature conditions which 
can be borne by the body rather than the optimal conditions. For 
example, it is possible to specify the conditions which will produce 
a dangerous burden upon the heart, which will disturb the water 


balance, or the like. There is, however, a wide range of temperature 
conditions within which the bodily changes are so small that we can 
attribute little importance to them. 

2. Effects upon health and disease. A number of investigators 
have studied the effects of weather and climate upon the rate of inci- 
dence of disease and upon the death rate.^^ It has been more difficult 
to obtain significant data upon the influence of indoor conditions. 
Here it is necessary to deal with relatively small experimental groups 
in a situation in which it is difficult to eliminate the influence of many 
other important variables, such as difference in exposure, differences 
in conditions outside the experimental room, and the like. While 
there are many theories about the influence of temperature conditions 
upon health, there is little well-founded information here. 

3. Influence of temperature conditions upon health, fertility, and 
activity of animals bred and raised under controlled conditions. 
While these studies serve to indicate that temperature is a factor 
worth considering, and may lead to the development of techniques 
which can later be applied to man, it is obvious that such observations 
could not establish optimal levels for man. As examples of this type 
of work we may refer to Mills' studies, in which he found that rats 
reared in a cool room (66° F. to 70° F.) grew faster and were more 
resistant to disease than those reared in temperatures of 89° F. to 
92° F.; 60-70 per cent RH.^^ 

4. Industrial performance and tested activities as influenced by 
temperature conditions. The Industrial Health Research Board of 
Great Britain has published a number of reports which show the influ- 
ence of seasonal temperature changes and different methods of ven- 
tilation upon output in industrial activity. For heavy work, the main 
problem is naturally to keep the temperatures low enough to permit 
easy loss of the excess heat developed in the body. 

For lighter activities, it has been difficult to discover the effects 
of temperature, if, indeed, they exist. The New York State Com- 
mission on Ventilation performed the most elaborate experiments upon 
this problem, using a variety of test activities, such as speed of nam- 
ing colors, cancellation, addition and multiplication, and typewriting, 
along with tasks requiring heavy muscular exertion. They were 

11 C. A. Mills, Medical Climatology, 1939. Ellsworth Huntington, Weather and 
Health, Bulletin of the National Research Council, 1930, No. 75, 1-161. 

12 C. A. Mills and C. Ogle, Control of body heat loss through radiant means, 
Transactions of the American Society of Heating and Ventilating Engineers, 1938, 
44, 167-178. 

13 Ventilation, Report of the N. Y. State Commission on Ventilation, 1923. 



[Ch. 6 

able to demonstrate that the influence upon muscular exertion was 
not a function of the ''freshness" of the air, but of temperature con- 
ditions alone. The ''mental" tasks, however, showed no effect of 
either temperature or freshness of the air. In fact, the subjects per- 
formed as well when the air was 86° F. and the humidity was 80 per 
cent as they did at 68° F. and 50 per cent R.H. When the subjects 
were left free to choose their own activity under the various condi- 
tions, there was no clear tendency to choose the easier tasks more 
frequently under warm conditions than there was under the more 
normal conditions. 

One possible interpretation of these results is that the subjects 
were "put on their mettle," by the experiment. Perhaps, then, they 
compensated for the effects of adverse temperature conditions by in- 
creased effort, just as they did under the influence of noise. Against 
this suggestion it might be argued that under the "normal" conditions 
they were supposedly working nearly as hard as their capacities to 
perform, so that any reduction in capacity when the temperature in- 
creased should have made itself evident in the test scores. The same 
argument, however, could just as well be applied to the experiments 
on noise, where, as we have seen, it is not borne out by the facts. It 
is likely that it would not apply to the effects of temperature either. 
Apparently it is not possible to obtain maximal performance on 
mental tests of this kind. There is always some additional effort pos- 
sible which may be brought out by adverse conditions of work — 
noise, or temperature, or others still to be discussed. 

Another possible interpretation is that temperature does not, in 
fact, influence mental activity very much. This is quite possible, 
since the bodily heat production involved in such tasks is so small. 
The only possible way to decide the question is by further experimenta- 
tion which takes effort as well as performance into account, using 
techniques similar to those employed for noise. This has not been 
carried out so far, and it is, therefore, impossible to determine stand- 
ards for temperature conditions upon the basis of performance until 
this has been done. 

5. Reports of comfort. It appears from the above summary of 
attempts to find some sensitive method of determining optimal tem- 
perature conditions that the only method now available for the pur- 
pose must be based upon the comfort of the subject, as perceived and 
reported by him. In many ways this also is subject to criticism, and 
we should prefer something better. We may choose temperatures 
which are too warm for our ultimate good just as we may eat more 



than is good for us, or too much or too Httle of certain foods. Nothing 
can be done about this, however, until the questions discussed above 
have been more adequately studied. It is possible, also, that rated 
feelings of warmth or cold are subject to unreliability, as are other 
ratings. They are probably affected by suggestion ; they will also be 
influenced by adaptation and acclimatization. 

In many studies in which the subjects report their degree of com- 
fort, there is no specification of the conditions from the point of 
view of radiation. The other three important factors are usually 
reported. In the examples shown in Table 10 most of the results are 
given in terms of ''Effective Temperature." It is necessary to under- 
stand this term before considering the results. The following is a 
rough statement of the meaning of the term. 

Effective Temperature. — A scale in wide use by heating and air- 
conditioning engineers, which shows the combined effect of air tem- 
perature and humidity for a specified rate of air movement. It was 
established by subjects who compared the warmth of two rooms of 
different humidity and temperature. Temperatures were adjusted 
until the rooms felt equally warm while the relative humidities re- 
mained fixed and different. Thus 66° F. effective temperature means 
that the conditions would feel the same as a room at 66° F. dry bulb 
temperature and 100 per cent relative humidity. Knowing the wet- 
bulb and dry-bulb temperatures in a room, it is possible to determine 
the effective temperature equivalent from a chart (Figure 16). In 
the examples given, air movement is always at a minimum. 

The results indicate that it is possible to suit a large proportion 
of the occupants of any building if their work is all of the same 
general sort, if seasonal variations in requirements are taken into 
account, and if variations in climate in various regions are also con- 
sidered. There are individual differences, of course, but the high 
percentages of subjects who are comfortable at the optimal level indi- 
cates that the amount of severe discomfort would be quite small. 

In establishing the optimal level in a given climate and for a given 
kind of activity, the votes should, of course, be obtained in a system- 
atic and controlled manner. If possible, it would be well to keep the 
subjects in ignorance as to the thermometer readings at the time the 
opinions are being taken, in order to prevent biases from entering the 
judgments. It would also be well to record the opinions after some 
period of time under a given condition. Those who have become 
accustomed to a given level may take some time to become accustomed 
to new conditions. 


so 60 70 60 90 iOO 


Note. — Both summer and winter comfort zones apply to inhabitants of the United States only. 
Application of winter comfort line is further limited to rooms heated by central station systems 
of the convection type. The line does not apply to rooms heated by radiant methods. Application 
of summer comfort line is limited to homes, offices and the like, where the occupants become 
fully adapted to the artificial air conditions. The line does not apply to theaters, department 
stores, and the like where the exposure is less than 3 hours. The optimum summer comfort 
line shown pertains to Pittsburgh and to other cities in the northern portion of the United States 
and Southern Canada, and at elevations not in excess of 1000 ft above sea level. An increase of 
one deg ET should be made approximately per 5 deg reduction in north latitude. 

Figure 16. American Society of Heating and Ventilating Engineers Comfort Chart 

for Still Air 

Effective temperature may be determined from dry-bulb temperature and relative humidity 
(or wet-bulb) reading. Effective temperature lines run from upper left to lower right. 

(From Heating Ventilating Air Conditioning Guide 1947. Reprinted by permission.) 



> <u 

C Ah 



O crt 

u t: 


-OX) > 
aj <u O 

O ui 


'^J- O 0\ *^ 

0\ 00 vO 00 

■5 o 2 

O tS c/2 i>4 

~ -^-^ O O 

C C 

o * 

U 00 t>. t-<; 

^ t-^ _<* ro 

Q VO fj- 

4, 00 00 


2 0. 

9. (U 




u •r 




E 2 

o ^ 


III til 

^ I o g - 


S 4i 

S 3 

! . ^"cii ^'w (J 
I p^" ^ ^ OS ^ .8 . 


^ §S .'2 c!^ 

Ej^ S^Er^^-S^W.Kr^ 



^. ■M . 8 . - 

•■S"^ oT* '^oT* 'i 

:^ ON ^ On 2- 

5^ -H 


Such votes are useful as a stopgap for use until other techniques 
can give us more satisfactory methods of determining optimal tem- 
perature levels. 

Lighting and Other Factors in Ocular Work 

Campaigns for ''better light, better sight," and the new develop- 
ments in lighting methods, have in recent years increased the attention 
which is given to the conditions of lighting for work such as reading, 
drafting, visual inspection, weaving, sewing, and many other similar 
activities. There are a number of interrelated questions involved in 
finding the optimal conditions for these tasks. There are problems 
of intensity, distribution and color of the illumination, and other 
factors, such as color of materials, which play a part. Intensity can- 
not be changed without producing some effects upon glare and dis- 
tribution of light, unless the lighting fixtures are also modified. 
Other interrelations hold for other combinations of these factors. 
While we shall consider these aspects separately in the following 
paragraphs, it is only an artificial separation, and the reader must keep 
these relationships in mind. 

Intensity of light. — Assuming optimal distribution and constant 
color of light, how bright should the light be for a given task? This 
question again raises the problem of the cost of work to the visual 
mechanism — the ocular fatigue involved. There are still differences 
of opinion concerning the correct levels of illumination to be adopted, 
due to the differences in evaluation of the measures of ocular fatigue. 

Two distinct points of view are represented by Tinker and Luck- 
iesh. Tinker points out that increases in illumination bring about 
increases in performance up to a certain point, as shown in Table 11. 
Beyond a certain level, however, no further advantage is gained by 
additional brightness of light.^^ Tinker has collected the results of a 
variety of studies to indicate the critical level for different tasks. 
These levels are shown in Table 12. Tinker does not believe there is 
sufficient evidence for any advantages to be gained by brightnesses 
very much above these values. 

Luckiesh, on the other hand, takes the position that eyestrain may 
be decreased by brighter lighting, even if there is no increase in per- 

19 M. A. Tinker, Illumination standards for effective and comfortable vision, 
Journal of Consulting Psychology, 1939, 3, 11-20. 

20 Tinker makes recommendations for practical lighting somewhat above the 
critical levels, e.g., about fifteen foot-candles for ordinary schoolroom tasks, allow- 
ing for a safety factor. See M. A. Tinker, Illumination standards, American Jour- 
nal of Public Health, 1946, 36, No. 9. 963-973. 



formance to indicate the benefit. It is conceivable, of course, that 
there are other Hmiting factors in ocular work which would establish 
a ceiling on performance and prevent indefinite improvement of out- 


Effects of Increased Illumination upon Industrial Output* 

Type of Initial Level New Level Per Cent of 

Factory (Foot-Candles) (Foot-Candles) Increase in 


Electrical 3.8 11.4 8.5 

Piston Ring 1.2 6.5 13.0 

" " 9.0 17.9 

" " 14.0 25.8 

Roller Bearing 5.0 6.0 4.0 

" 13.0 8.0 

20.0 12.5 

* After Troland (table assembled by Tinker, op. cit., 1939, p. 17): To present a con- 
crete idea of the meaning of the foot-candle unit of illumination, a 100-watt bulb in a 
bridge-lamp shade will give 10 foot-candles at a distance of 52 inches, and 2 5 foot-candles 
at 34 inches. There is, however, variation with the size and color of shade, brightness of 
walls, and the like. 


Critical Levels of Illumination f 
Beyond Which There Is Little Change in Output 

Task Critical Level 


Reading 10-point type (speed) 3 

Reading- 10-point type (fatigue, blurring) 3 

Reading 12-point type (speed) Between 1 and 10 

Reading performance (school) 4-6 

Educ. achievement in school 4-6 

Computing arithmetic problems Less than 9.6 

Sorting mail 8-10 

Setting 6-point type by hand 13-25 

Time to thread a needle 30 

t Tinker, op. cit., 1939, p. 13. 

put. Thus, in reading, we cannot read faster than we are able to 
comprehend. Luckiesh's argument would be that output alone cannot 
be used to establish the optimal levels of illumination, and this seems 
a reasonable argument. The final decision depends, however, upon 
our ability to measure fatigue in ocular work by methods independent 
of the output. 



[Ch. 6 

In a series of studies in the General Electric Company laboratories 
at Nela Park, Luckiesh and his collaborators have investigated the 
effect of illumination upon a number of proposed measures of ocular 
fatigue. The choice of these measures is based upon the following 
reasoning: (1) these tests involve the ocular mechanisms (accommo- 
dation, etc.) or other mechanisms which are involved in the total 
activity of reading {e.g., muscular tension) and (2) these measures 
change under conditions which are obviously bad (the subject wears 
incorrect lenses or works under a very bad glare) . 

A summary of the effect of intensity upon these measures is given 
in Table 13. 

TABLE 13* 
Influence of Level of Illumination 

Type of Test Intensity of Illumination (Foot- Candles) 

1 10 100 

Visual acuity 100 per cent 130 per cent 170 per cent 

Contrast sensitivity 100 per cent 280 per cent 450 per cent 

Muscular tension (Key pressure) 
vv^hile reading 63 grams 54 grams 43 grams 

Frequency of blinking (Change 

after reading one hour) 100 per cent 77 per cent 65 per cent 

Decrease in heart rate while read- 
ing one hour 10 per cent 2 per cent 

Decrease in convergence reserve of 
ocular muscles after reading one 

hour 23 per cent 7 per cent 

* M. Luckiesh, Light, Vision, and Seeing, New York, D. Van Nostrand Co., Inc., 1944. 
p. 200. 

It will be noted that two of the measures listed in Table 13 are 
measures of capacity for discrimination (acuity and contrast sensi- 
tivity). They do not indicate the cost of performing normal tasks 
where the discriminations required are far above the limits of vision, 
unless we make the assumption that all work is made easier if the 
limits of .performance are raised. Such an assumption has been 
questioned, however, so that the effect of illumination upon capacity 
alone cannot be conclusive in settling the problem. Other measures 
of capacity have also shown similar effects of raising the intensity of 
illumination. Ferree and Rand, for example, have found increased 
speed of discrimination and speed of focusing, the speeds still increas- 
ing up to 100 foot-candles.^^ In all cases, acuity or speed increase 

21 C. E. Ferree and G. Rand, Intensity of light and speed of vision studied with 
special reference to industrial situations, Transactions of the Illuminating Engi- 
neering Society, 1927, 22, 79-110. 



rapidly as the light passes from zero through the lower ranges of in- 
tensities, and more and more slowly as higher levels of intensity are 
reached. In Luckiesh's data the increases are proportional to the 
ratio of increase of light rather than to the absolute increase. Thus 
the gain from additional light energy is relatively small, once we pass 
beyond 10-20 foot-candles. Since this phenomenon of diminishing 
returns applies to the other measures of cost, it is used by Tinker as 
an additional argument against the desirability of high intensities. 

The other measures (tension, heart rate, convergence reserve, 
blinking) are considered as indices of the strain which ocular work 
puts upon the organism as a whole or upon the ocular musculature. 
These results have been questioned by Tinker and others, not only 
because of the diminishing returns already mentioned, but also be- 
cause the measures are regarded as of doubtful validity or reliability. 
Space will not permit an analysis of this question here, but it appears 
that at least the rate of blinking and of the heartbeat have little value 
as indicators of fatigue in ocular work.^^ In addition, it should be 
mentioned that Tinker has experimented with a further test of fatigue 
which Ferree and Rand applied with success to problems of light dis- 
tribution and color. Tinker could find no change in this index be- 
yond about 3 foot-candles.^^ 

Ferree and Rand approached the question of intensity of light from 
the point of view of individual differences. They had found that 
additional light has more marked advantages for the older person who 
has lost part of his power of accommodation.^* Acuity showed 
greater increases for these subjects than it did for younger, normal 
subjects. Their later studies of individual differences related to 
preferences in light rather than to functional tests. Since these 
preferences are likely to be a function of habits and customs of the 
subjects, it is not likely that they will solve the problem before us.^^ 

22 R. A. McFarland, C. A. Knehr, and C. Berens, Metabolism and pulse rate 
as related to reading under high and low levels of illumination, Journal of Experi- 
mental Psychology, 1939, 25, 65-75. R. A. McFarland, A. H. Holway, and L. M. 
Hurvich, Studies of Visual Fatigue, 1942 (Boston). M. E. Bitterman, Heart rate 
and frequency of blinking as indices of visual efficiency, Journal of Experimental 
Psychology, 1945, 35, 279-292. 

2 3 M. A. Tinker, The effect of illumination intensities upon the speed of percep- 
tion and upon fatigue in reading. Journal of Educatio}ial Psychology, 1939, 30, 

2* C. E. Ferree and G. Rand, The effect of increase of intensity of light on the 
visual acuity of presbyopic and non-presbyopic eyes, Transactions of the Illuminat- 
ing Engineering Society, 1934, 19, 296-313. 

2 5 C. E. Ferree and G. Rand, Good working conditions for eyes, Personnel 
Journal, 1937, 15, 333-340. 

2 6 Tinker has stressed the importance of the adaptation of the subject in evaluat- 
ing the results of a preference study (M. A. Tinker, Effect of visual adaptation 



[Ch. 6 

Of course the same might be said of preferences for temperature con- 
ditions, but here we must recognize the existence of ocular strain as 
the central problem to be solved. In the field of temperature there is 
no clearly recognized strain due to temperature unless the conditions 
are extreme. 

To conclude, there is little indication in this set of contradictory 
findings that any great benefits are to be expected from levels of light 
above 10 foot-candles for ordinary reading, with somewhat higher 
levels for more exacting tasks such as fine sewing or inspection work. 
With modern lamps and lighting fixtures, there is probably little dan- 
ger of inadequate intensity of light, although that was not the condi- 
tion a few years ago. In special tasks, however, greater attention to 
this problem may still be necessary. There is also no evidence that 
high intensities are harmful, provided that the factor of glare is 

Glare and Distribution of Light.— Glare is probably a more im- 
portant factor in ocular efficiency than intensity of light, at least under 
average conditions. We can discuss glare more briefly, however, 
since there is no disagreement about the problem. Bright spots in 
the field of vision make the seeing of details more difficult, and all the 
various indices of ocular fatigue show the effect of these conditions. 
This is the general principle, which has a number of implications. 

upon intensity of illumination preferred for reading with direct lighting, Journal 
of Applied Psychology, 1945, 29, 471-476). His experiment, dealing with localized 
lighting, did not, however, indicate any marked effect of the adaptation level upon 
choice. The results were about the same whether the subject was adapted to 
twenty or to fifty foot-candles before making his choice. When choosing more 
generally distributed illumination, however. Tinker found a more marked effect 
of the adaptation level. 

The adaptation to which Tinker refers includes only one factor which may play 
a part in determining preferences for intensity levels. In addition to this factor in 
the immediate history of the ocular mechanism, there is probably a more general 
kind of "growing accustomed" to certain general levels of lighting for reading or 
other ocular tasks. This factor could be controlled only by finding out the habits 
of the subjects, the type of lamp which they ordinarily use, the customary position 
when reading, and the like. If sufficient data of this kind were collected, then the 
factor of habituation might be controlled by statistical methods. 

2 7 The measures of fatigue and capacity which have been used in the study of 
glare are the same as those employed in studying the effect of intensity. Reliability 
and validity are no greater in one case than in the other, but since there is greater 
agreement among various measures and different investigators, the importance of 
the factor of glare is usually accepted without question. For example, Ferree and 
Rand's li test, which showed no advantage of high levels of intensity, is reported 
to show very consistent effects of bright light sources in the field of vision. 

It should be noted, however, that until these measures are more fully established 
and validated, we cannot say how serious it is to have a given bright source in the 
field of vision. We can be sure it has some harmful effect, but how great this 
effect would be in comparison with other effects, such as the fatigue due to reading 
very fine print, is very difficult to say. 



Lighting fixtures should not be visible when the subject is work- 
ing. Shading the bulbs is not enough, because the shade rim itself 
will be a bright spot contrasting with surrounding darker objects. 
Therefore the fixtures should be removed from the field of vision 
altogether. "Indirect" systems of lighting are not always satisfac- 
tory for this purpose, because the ceiling or walls which reflect the 
light may themselves become bright spots if sufficient intensity is sup- 
plied at the working place. It should also be noted that bright spots 
may be produced by highly polished objects as well as by lighting 

While there should be no bright areas in the field around the 
work, there should also be no very dark areas. If a part of a room 
is very dark, lighter objects will contrast with the dark areas and 
produce the effects of still brighter objects. For these reasons it is 
usually recommended that the room be supplied with a moderate level 
of even illumination, with extra light falling upon the work from fix- 
tures which are above and behind the worker. 

Color of Illumination. — Among monochromatic (single wave- 
length) lights, the eye is more sensitive to the middle or yellow part 
of the spectrum. Even when the illumination is constant in apparent 
brightness, and varies only in hue, Ferree and Rand found that 
acuity was greater under yellow or amber light than it was under 
any other colored light. They also found a difference in fatigue (li 
test) produced by reading under the various colored lights.^^ Among 
mixed lights, it is found that daylight is more effective than tungsten 
lamps. It must be remembered, however, that filters for tungsten 
lamps which duplicate the spectral distribution of sunlight also reduce 
the intensity of the light. This means, of course, a greater expense 
in maintaining the level of intensity required. Ferree and Rand 
report that some so-called "daylight" filters do not produce any ad- 
vantage over ordinary tungsten lamps. 

Working Materials and Conditions. — Ease of seeing is a function 
of the amount of contrast between details to be discriminated and 
their background. This contrast is enhanced by adequate illumina- 
tion, but it can also be controlled by the manufacturer of books, the 
designer of equipment, the painter of machinery. Paper should be 
as white as possible, and have a surface which does not produce mir- 
ror-like reflection. Newspapers and catalogues are often fatiguing 

2 8 C. E. Ferree and G. Rand, Visibility of objects as affected by color and com- 
position of light: Part I, Personnel Journal, 1931, 9, 475-492; Part II. ibid., 1931, 
10, 108-124. 



[Ch. 6 

because of the quality of the paper. Dark parts or tools can be seen 
more, readily if they are provided with a light background. It has 
been found helpful in some industrial operations to work on a ground 
glass plate through which diffuse light shines from below. 

These are a few examples of the many devices which may increase 
ease of discrimination without raising the level of illumination, and 
often at very small expense. Research is also providing book manu- 
facturers with much information upon type faces, margins, leading, 
etc., all of which affect the ease of reading. On some points there is 
controversy, due to the same difficulties which were met in deter- 
mining optimal levels of illumination. Nevertheless there are enough 
facts established to provide a sound beginning. Until measures of 
fatigue are more adequately validated, it would probably be best to 
follow the suggestions of Paterson and Tinker which are based upon 
performance tests.^^ 

There are many other problems of visibility which can be settled 
only by special psychological experiments. The optimum arrange- 
ment and design of automobile headlights to provide maximum dis- 
tance of perception at night, the placement and design of highway 
signs, and other similar information, are of most immediate interest 
to the engineer or highway official, since the general public cannot put 
such information into practice. The applied psychologist needs to be 
aware of the existence of studies of this kind and be prepared to rec- 
ommend that specialists in this kind of work be called upon to study 
new problems as they arise. 

As one example of this kind of work, we may refer to the recently 
published work of Berger upon the visibility of numbers. He has 
studied the shape and proportion of numerals, and has designed a set 
which are more visible than those now used on license plates. In ad- 
dition, all the numbers are equally visible. In this analysis, the effects 
of width of line, spacing of numbers, black on white versus white 
on black, the shape, and area of the numbers were all taken into 
account. In this manner it was possible to produce numerals of the 
same size which were about 36 per cent more legible than previous 
numerals which were used for comparison. White numerals on a 
black background were more easily recognized than black numerals on 
a white background, and slender luminous numerals were more effec- 
tive at night than the numerals seen under reflected light. 

2 9 D. G. Paterson and M. A. Tinker, How to Make Type Readable, New York, 
Harper & Bros., 1940. 

3 C. Berger, Stroke width, form, and horizontal spacing of numerals as deter- 
minants of the threshold of recognition, Journal of Applied Psychology, 1944, 28, 
208-231, 336-346. 



Hours of Work, Rest Periods, and Sleep 

Some of the effects of these factors have appeared in laboratory 
studies of fatigue and efficiency, and have already received brief men- 
tion in this book. Their more general effects in the world of work 
outside the laboratory must now be surveyed. The scheduling of 
working time, which involves all the factors listed in the title of this 
section, has come in for considerable discussion and some research. 
It is a topic which becomes especially important during periods of 
high production requirements. Even in slack seasons when the labor 
supply is plentiful, there are problems of this kind which require re- 
search. For example, is it more efficient to employ many workers 
for very short periods, or fewer workers for longer periods ? 

The effects of the length of the working day, of the weekly hours 
of work, of formal programs of rest periods in industrial production, 
have all been investigated primarily in terms of their effect upon total 
production, with an occasional reference to ''morale," turnover, and 
accident frequency. The effects of these factors upon efficiency, in 
our sense of the term, have either been altogether overlooked or de- 
duced by commonsense analysis from production data. Some of the 
recommendations for optimal conditions, particularly in regard to 
rest periods, have not been based upon direct study in industry, but 
have been drawn from laboratory studies of work on the ergograph, 
or some other task whose resemblance to normal working activities is 
extremely remote. 

We shall not have space for a detailed examination of the research 
on these topics. We shall try only to indicate its general character, 
and give some of the main conclusions which can be drawn from the 
group of researches now available. 

Hours of Work. — Problems of wartime production during the 
first World War gave impetus to the study of optimal hours of work 
from the point of view of obtaining maximum total output per 
worker. During that war the British established a Health of Muni- 
tions Workers Committee which investigated the problem. Later, 
the Industrial Health Research Board continued the analysis, and 
drew conclusions which have been supported by more recent investi- 
gators. Most of these results are summarized by Vernon. 

Several of these studies were begun at a time when the weekly 
shifts in munitions factories were extremely long (often over seventy 

31 H. M. Vernon, Industrial Fatigue and Efficiency, New York, E. P. Button 
& Co., 1921, pp. viii, 263. 



[Ch. 6 

hours per week). It was found that a reduction of the weekly hours 
increased the average hourly rate of production. Often this increase 
in rate of work was so great that it more than balanced the decrease 
in number of hours worked. The effect was not so marked in those 
jobs where the rate of work was largely controlled by automatic ma- 
chines. Even here, however, the loss in total output was not so great 
as the reduction in number of hours worked. This phenomenon is 
due to the fact that total output is a function of the actual working 
time of the individual (nominal time minus rest periods, interrup- 
tions, etc. ) . A smaller proportion of working time is lost from these 
interruptions in the shorter shifts. 

The optimal length of a working week cannot be stated in gen- 
eral terms. As indicated above, the optimum, from the point of view 
of total production, varies with the nature of the job. In some of 
the jobs reported by Vernon, the largest output occurred at fifty-five 
hours (the lowest studied), in others the optimum was at the highest 
number of hours, or over seventy. It must be remembered that most 
studies of this kind involve the special motivational conditions of 
wartime. There is no guarantee that the same optimum would be 
found in peacetime, even for the same kind of work. The only way 
in which the optimum for a given job can be determined is by long- 
term trial. 

Another difficulty with attempts to state optimal hours of work 
is that the optimum, from the point of view of total production, is not 
necessarily the optimum from other points of view. In some ways we 
might consider the average hourly rate of work as a better indicator 
of the efficiency of the individual worker. The hourly rate also 
shows variation with varying length of shifts, going down when the 
shifts are either very long or very short. The highest rate of work 
often appears, however, in a shift which is shorter than that which 
brings maximum total production. 

As we have suggested before, however, the rate of work alone is 
no more satisfactory as an indicator of efficiency than is the total out- 
put. We might work at maximum speed if we worked only one hour 
per day. Nevertheless, total output must be given some considera- 
tion. Our ideal is a maximum ratio between the total output per 
week and the total cost of the work for the week. This point is prob- 
ably not the point where we attain the maximum total output, nor is it 
the point where we achieve a maximum hourly rate of work. A 
point somewhere between would be the most likely optimum. At 
present, however, we have no information which would permit us to 
determine that optimal point for the hours of work, even for a 



single job. Until information of that kind can be secured, and in 
times of pressure for high production, the point of maximum total 
output would appear to be the best choice. 

The Industrial Fatigue Board studies provide some indirect clues 
as to the effect of changing hours of work upon efficiency of per- 
formance. When the length of the working week is reduced, it may 
take as long as eight months for the rate of production to reach a new 
equilibrium. When the week is first shortened, production may go 
along for some time with only slight increases in hourly rate. Then 
it gradually moves up to a new level. In some instances the first rise 
in production rate may overshoot its ultimate level for a time, with 
a subsequent reduction to the new stabilized level. 

This gradual process of adjustment of the worker to the new 
schedule might be interpreted to mean that the worker, intentionally 
or unintentionally, divides up the quota of effort which he is willing 
to expend on the job. He spreads that quota over the working week. 
When the working week is shortened, he has to learn how to re-divide 
his quota of effort, gradually increasing the allotment to each hour 
and day until he is again expending the same total which he used 
before. This is a rough statement of the interpretation which Vernon 
gives. If it is true, it would mean that an increase in total production 
means an actual increase in efficiency. Since the cost is adjusted at 
approximately the same level as before, the increase in output is pure 

To follow the logic of this interpretation still further, we may 
assume that the worker is also limited in the rate at which he can 
expend effort at any one time. As the hours of work are shortened 
below the optimal point, he has more and more difficulty in expend- 
ing the quota of effort which he is willing to give to the work. Con- 
sequently the total output goes down. 

When the hours of work are increased, however, the process of 
adjustment is no longer gradual. There is an immediate reduction 
in hourly rate of work, although here again there may be an over- 
shooting, with some gradual readjustment later. It would appear 
that the mechanism of this adjustment is different from, and prob- 
ably more deliberate than, that which accompanies reductions in the 
hours of work. 

We could probably build still further upon Vernon's theory of a 
quota of effort (which Vernon calls ''energy"). For example, the 
effect of changing motivation might be described in terms of a re- 
duced quota of effort which the worker is willing to devote to the 
job. In fact we shall make use of this conception in our later chap- 



[Ch. 6 

ters upon the special topic of motivation. Before building too far, 
however, we must remind ourselves that the whole theory is based 
upon a supposition which needs to be checked experimentally. This 
represents one of the important tasks awaiting the development of 
more complete indices of effort and the cost of work. 

Another, more immediate, implication of the gradual adjustment 
to reduced hours of work concerns the practical problem of finding 
the optimal hours of work for a given job in terms of total output. 
Since there is no single optimum, as we have indicated before, each 
plant or industry must determine its own recommendations. These 
recommendations must not, however, be based upon brief periods of 
observation. Vernon's studies indicate that at least eight to twelve 
months must elapse in testing each length of working period. We 
might also suggest that motivational conditions should remain at 
least approximately constant during that time if the results are to be 

The slow adjusment to decreased hours and the rapid adjustment 
to increased hours indicates, according to Vernon, a serious disad- 
vantage of irregular overtime work. He believes that a worker who 
is called upon occasionally to work for a ten-hour shift in place of a 
regular eight-hour day, is likely to work at a rate adapted to the ten- 
hour period, not only on the days when he knows he is to work 
longer, but all the time. For this reason it would be better to spread 
the overtime evenly throughout the week, rather than concentrate it 
in one or two days. 

The comparative effects of night and day shifts have been dis- 
cussed at some length on a common-sense basis. Very little accurate 
information is available, however. In many plants the workers on 
the night shift are not equivalent in skill to those on the day shift, 
since seniority may give a worker a right to choose daytime work. 
In some instances the supervision is also less adequate at night. For 
these reasons, comparison of production records would have little 
meaning. The answer to the question of whether a night shift is 
less efficient than a day shift is largely academic, in any event. When 
workers are employed on a night shift, they are employed because 
the plant must be kept in operation day and night, and there is no 
choice involved. A more pertinent question, therefore, would be : 
What factors in efficiency differ in the day and night shifts, and how 
can they be controlled in such a way as to increase the effectiveness 
of both? 

Whenever factories are working ''around the clock," there is one 
choice which can be made. This is the choice between assigning 



workers permanently to a given shift, or using a system of rotation. 
The major advantage of the rotating shift is that it is fairer to all 
concerned, since few prefer to work at night. On the other hand, 
it takes some time for an individual to readjust his habits for night 
work. He may not be able to complete the adjustment before it be- 
comes time to change to new working hours. 

Vernon has also reported statistics bearing upon this problem. 
The output of a group of workers on the night shift differed little from 
the output of the same workers when working in the daytime. This 
finding is not conclusive, according to Vernon, because deleterious 
effects of the night work might carry over into the succeeding period 
of day work, the periods of alternation being rather short. In other 
cases he compared the output of workers who were alternating shifts, 
with their later output when they were working continuously in either 
a day shift or a night shift. The output on the continuous day shift 
was only slightly better than that for the alternating shifts. The 
group which changed to the continuous night shift, however, declined 
about 4 per cent in output. 

Thus there seems to be but little evidence of any marked effect of 
the arrangement of shifts upon the output of the workers. The most 
important factor in the choice would therefore be the preference of 
the workers as a group — the choice of the arrangement which seems 
fair to the greatest number. 

Rest Periods — The basic question in a discussion of rest periods 
in industry does not involve a choice between the presence or absence 
of rest periods. The studies of Vernon and others indicate clearly 
that some form of rest period is taken by the worker even where there 
is no formal plan allowing for them. The problem consists rather of 
determining whether or not a formally authorized rest period is more 
effective than an unauthorized rest, and if so, what program of rest 
is most beneficial. 

As in the case of hours of work, the principal criteria of the value 
of rest periods have been hourly rate and total output per day. The 
results of introducing regular rest periods vary from slight increases 
in hourly rate which are insufficient to offset the time lost in the rest 
period, to increases which are sufficient to increase the total output by 
substantial amounts. There are enough data to warrant the conclu- 
sion that nothing is lost by the use of regular rest periods, and con- 
siderable gain in output and "morale" may result.^ 

It is to be expected that some arrangements of rest periods would 
be more effective than others, but there have been few experiments in 



[Ch. 6 

which the schedule of rest periods has been varied systematically in an 
industrial setting. This is understandable when we consider that it 
may take a long time for the full effects of the rest periods to be 
reliably established. 

/ The laboratory studies of rest periods had indicated similar gen- 
eral principles for both muscular and mental work. Graf, for 
example, performed experiments upon rest periods in continuous 
arithmetical calculation (three-hour periods). He found that a given 
amount of time spent in rest was more. effective if the rests were brief 
and frequent. In different series of experiments, a total of twelve 
minutes of the three-hour period was divided into anywhere from two 
to eleven rest periods. In general, the larger number of short rest 
periods led to the greatest gains. Graf also found that the results 
were better if the length of the rest periods was gradually increased 
from the beginning to the end of the work-spell. 

Even though we do not have very much direct confirmation from 
industrial studies, there seems to be general agreement that these 
principles derived from laboratory studies are applicable to everyday 
tasks. One instance of a systematic variation in rest periods is Rich- 
ter's study of women working in a metal working plant, which shows 
a pattern similar to that obtained by Graf for a laboratory task.^^ 

One of the most elaborate studies of the effect of rest periods will 
not be analyzed in detail in this chapter because the results have more 
bearing upon the problem of motivation than they have upon the ef- 
fects of rest periods as governing factors in the efficiency of work. 
These studies, generally known as the ''Western Electric" or ''Haw- 
thorne" experiments, began as studies of the effects of rest periods but 
resulted in few conclusions which have a bearing upon that problem.^* 
The experiments failed to control the motivational factors which 
must be held reasonably constant in an adequate study of the present 
problem. The motivational factors loomed so large that the investi- 
gators became more interested in studying them. 

As our later analysis of the Hawthorne research will indicate, 
there is nothing in these results which contradicts the opinion that 
rest periods are beneficial, although some readers have mistakenly 
drawn this conclusion. In fact, that company proceeded to install rest 

3 2 0. Graf, tJber die Wirkung mehrfacher Arbeitspausen bei geistiger Arbeit, 
Kraepelin's Psychologische Arheiten, 1925, 9, 1-69, and Die Arbeitspause in Theorie 
und Praxis, i&td, 1927, 9, 563-681. 

3 3 W. Richter, Leistungssteigerungen in der Blankschraubenfabrikation durch 
Einfiihrung von Zwangspausen, Industrielle Psychotcchnik, 1931, 8, 129-146. 

34 F. J. Roethlisberger and W. J. Dickson, Management and the Worker, Cam- 
bridge, Harvard University Press, 1939. 


periods in other departments even though they could not draw con- 
clusions on their effects from the experimental studies. 

It is true that rest periods often have as much (or more) effect 
upon the motivation of the worker as they do upon the relief of 
fatigue. Motivation itself is a factor in efficiency, however, although 
the way in which it affects efficiency may be somewhat different from 
the effects of other factors. Even if the effect of rest periods were 
due wholly to increased motivation of the worker rather than to relief 
of fatigue, there is still a problem of efficiency to be considered. It 
is true that most of our knowledge of the effect of rest periods con- 
cerns output rather than efficiency. This indicates a gap in our 
knowledge which must be filled. Until it is, we may assume that the 
increases in productivity of the worker under rest periods are a reflec- 
tion of increased efficiency. 

There have been occasional studies of mechanical efficiency as a 
function of rest periods. In one of these, concerned with machine 
operators in a foundry, productivity and efficiency were both increased 
by rest periods. In other studies, output decrement has been re- 
duced by the introduction of rest periods. The work curves of Graf, 
for intellectual work in the laboratory,^® and of Richter for factory 
work^^ both indicate this flattening of the output curve with optimal 
arrangement of rest periods. 

This information upon mechanical efficiency and upon work decre- 
ment covers only part of the factors involved in total efficiency, and 
only a few specific jobs. We also have, however, the fact that output 
increases on hourly paid jobs when rest periods are introduced. This 
should give us additional confidence in the efficiency of formal rest 

Sleep. — The topic of sleep involves questions which are usually the 
personal concern of an individual rather than problems which prop- 
erly belong in the field of industrial psychology. Nevertheless, in 
certain occupations, sleep needs become a serious concern and a matter 
of policy to the organization as a whole as well as to the individual 
worker. Such occupations are aviation, transport in general, and 

3 5 A. S. Borschtschewski, S. D. Gorkin, and B. J. Kustanowitsch, Rational- 
isierung des Arbeitsregimes beim maschinellen Formengiessen, Arbeitphysioloqie , 
1932-1933, 6, 311-338. 

3 6 O. Graf, op. cit. 

3 7 W. Richter, op. cit. 

3 8 Little has been said about the statistical reliability of the results cited in this 
section. In most cases this is not given, and the investigators do not appear to 
recognize this need. Nevertheless there is sufficient uniformity of results to justify 
some confidence in the general conclusions which we have drawn. 


[Ch. 6 

many military occupations. Even if it does have some importance in 
certain phases of industrial psychology, however, the topic of sleep 
is so complex that we shall attempt to give no more than a brief sum- 
mary of its present status. 

Here, as in many other phases of applied science, the solution of 
the immediate practical problems awaits the solution of more funda- 
mental problems. The immediate problems are, of course, such 
questions as these: *'How much sleep does an individual need?" 
"How long can one go without sleep without endangering health or 
safety in operating a car or machine ?" ''How can the depth or 'good- 
ness' of sleep be improved?" "How can insomnia be prevented or 

These questions cannot be answered, however, until we know 
more about the nature of sleep itself, what function it performs for 
the organism, and what factors affect it. In order to develop reliable 
experimental techniques which would arrive at knowledge upon these 
fundamental problems, it is necessary to develop some adequate cri- 
teria for determining when an individual is asleep, and how deep his 
sleep is at any given time. There are so many borderline states 
between sleep and waking, and an individual's own reports upon sleep 
are so untrustworthy, that the development of adequate criteria is an 
essential step in the study. 

As an example showing the importance of criteria of sleep, we may 
mention the question of the distribution of sleep during the hours of 
the night. It is frequently claimed that we sleep more deeply during 
the early hours than we do toward morning. There are a great many 
bodily functions which undergo changes correlated with sleep. Some 
of these, however, show the greatest effect during the early hours, 
while others show the greatest effect later on. Since we have no way 
at present of saying which criteria are more intimately related to the 
depth or soundness of sleep, we can take our choice of criteria and 
prove that the soundest sleep occurs at any period we wish. We shall 
not go into the details of these bodily functions since they are re- 
viewed with considerable completeness in Kleitman's book on the 
subject. At any rate, there is no evidence to support the super- 
stition that "we do most of our sleeping" in a few hours, and that the 
remainder of the time is wasted. 

We cannot state definite sleep requirements either for individuals 
or as group averages because there is as yet no way of measuring the 

3 9 N. Kleitman, Sleep and Wakefulness, Chicago, University of Chicago Press, 



effects of loss of sleep in any meaningful way. In terms of quanti- 
fiable measures, even prolonged loss of sleep (three or four days) 
produces surprisingly little effect. Kleitman found that physiological 
measures involving circulation, excretion, and the like, showed almost 
no change at all. Psychological tests also showed little effect, unless 
the test called for prolonged effort and continuous attention during a 
considerable period of time. For brief periods, performance was just 
as good as that in control periods. 

Johnson has stressed the point of view that the nature of these 
results is a function of the tests which are used.*^ In order to get 
reliable and quantifiable tests, the tasks have to be rather simple and 
routine in character. Subjects who have been without sleep for pro- 
longed periods may show many disturbances, such as hallucinations, 
irritability, delusions, and the like. Yet they perform well on the 
routine tasks in the tests. These findings are very similar to those 
of experiments upon the effects of alcohol, and so Johnson suggests 
that sleep deprivation has effects very similar to those of a narcotic. 
Disturbances very evidently occur, but their maximum effect involves 
functions which have not yet been adequately brought under the con- 
trol of psychological tests. 

This conclusion of Johnson's has a significance beyond the topic 
of sleep alone. Evidently tests which cannot detect the effects of such 
drastic conditions as prolonged loss of sleep must also miss the effects 
of fatigue when they are used as ''fatigue tests" to measure the cost 
of work. Even where it is possible to show that a given test-decre- 
ment is correlated with the duration of work, it probably does not 
reflect all the fatigue. Until we have measures which indicate the 
performance of the individual in the more complex activities involv- 
ing thinking, imagining, comprehension, and emotion, we are bound 
to miss many of the elements of cost of work when we use the tech- 
nique of fatigue tests. 

It must also be pointed out that tests designed to reflect the effects 
of sleep-loss reflect changes in capacity rather than changes in efli- 
ciency.*^ Not only that, but they may not be completely valid indi- 
cators of capacity because of the subject's tendency to leave some 
"cushion" or "safety factor" even under the high motivation of a 
laboratory experiment. The lack of sleep may have the effect of re- 
ducing capacity on a given test, but at the same time the gap between 

40 H. M. Johnson, T. H. Swan, and G. E. Weigand, Sleep, Psychological Bul- 
letin, 1926, 23, 482-503. 

*i See discussion of capacity tests in Chapter 4. 



[Ch. 6 

actual performance and capacity is reduced. Performance, therefore, 
remains apparently little affected by the sleep-loss. 

The direct measurement of efficiency of performance following 
sleep-loss would appear to be a more satisfactory approach to the 
problem than that which depends upon tests of capacity alone. This 
is, of course, more easily said than done. For this reason, and also 
because many investigators do not appear to recognize the need, few 
studies of the cost of work following sleep are available. Laird and 
Wheeler have reported a study of oxygen consumption during mental 
calculation.*^ They reduced the sleep of their subjects from eight 
to six hours per night for a period of one week. Although the sub- 
jects did not decrease their performance (errors remained the same, 
and speed increased), the oxygen consumption was increased during 
the "insomnia" period. This study does not have much significance, 
however, for the following reasons : ( 1 ) Although the results w^re 
marked and consistent, there were only three subjects. (2) As we 
have already seen, oxygen consumption is not a reliable indicator of 
the cost of mental work. (3) Other studies of Laird's showing the 
effect of mental activity upon oxygen consumption have not been 
corroborated by other investigators. We mention the study only to 
show that this general approach has been attempted, and also because 
the value of the study as an investigation of efficiency is sometimes 
overestimated in writings upon the subject. 

It will be seen that the attack upon the fundamental problems of 
the nature and effects of sleep has done little more than clarify the 
problem and show up some of the pitfalls which await investigators in 
this field. In the light of this situation, it is understandable if an 
investigator who wants to achieve results in a practical study of sleep 
problems goes back to more qualitative methods, and to a dependence 
upon the reports and comments of his subjects. This is essentially 
what Kleitman has done in an experiment upon the effects of certain 
foods taken before retiring.*^ 

In part of the study, Kleitman recorded movements during sleep 
as an indicator of depth of sleep, but the bulk of his conclusions was 
based upon a questionnaire which was answered by the subjects each 
day. The subject reported upon the duration of sleep, the presence 
or absence of dreaming, the "quality" of the sleep, whether or not 
he felt refreshed when he awakened, and upon incidents of the day 

4 2 D. A. Laird and W. L. Wheeler, Jr., What it costs to lose sleep, Industrial 
Psychology, 1926, 1, 694-696. 

4 3 N. Kleitman, F. J. Mullin, N. R. Cooperman, and S. Titelbaum, Sleep Char- 
acteristics, Chicago, University of Chicago Press, 1937. 


which may have affected these factors. Various substances, including 
a popularly advertised ''bedtime" food-drink, were administered on 
experimental nights before retiring. Other nights served as con- 
trols. In addition to the effects of the foods taken before sleeping, 
the effects of weather and season were also analyzed. 

All the data showed wide variations from subject to subject, and 
for the same subject on different nights. Whether this variation is 
due to the unreliability of the method of obtaining the information, 
or to valid changes in these variables, is not, of course, known. At 
any rate, the variation was so great that few significant effects of the 
controlled variables could be discovered. The authors conclude that 
the patent food ''significantly decreased the motility during sleep," 
and that the patent food "is the only material which, when taken 
prior to going to bed, increases the percentage of mornings on which 
one awakens well rested." In light of the data presented by these 
authors, these statements would seem incautious, especially since they 
provide the advertisers with ammunition for a high-powered sales 
campaign. The levels of significance which were accepted were not 
very stringent for conclusions which are to be used in this manner, 
and there is not even internal consistency of the results. It was true 
that the patent food resulted in significantly more reports of "feeling 
well rested" when the nights in which this food was used were com- 
pared with the control nights (when nothing was administered) . The 
effects of the food were not significantly different from the results 
for nights in which water or milk was given without the food. In 
fact, the effect of cold water upon this index was almost as significant 
when the comparison was made with the control nights. 

There are similar inconsistencies in the results with the objective 
measure of motility during sleep. The patent food was significantly 
effective when administered in cold water or hot milk, but not when 
it was given in hot water or cold milk. In fact the probabilities vary 
all the way from .90 down to less than .01, depending upon the 
medium in which the food was administered. In the case of motility, 
however, there was a moderately significant difference between the 
group of all nights when the patent food was administered, and the 
nights when milk or water was administered alone. 

The difficulty of drawing significant conclusions from studies 
based upon such qualitative indices is not only in the extreme varia- 
bility of results, but also in the fact that we have no way of deter- 
mining what meaning they have for the general efficiency of the 

44 Kleitman, Mullin, Cooperman, and Titelbaum, op. cit., p. 85. 



[Ch. 6 

individual. Of course, if an individual feels refreshed after sleeping, 
that in itself is a gain. But does this fact mean anything more than 
that? Is there any relation to his continued efficiency through the 
day? Until the answers to such questions as this are known, we 
could not interpret Kleitman's results, even if they were highly con- 
sistent and statistically significant. 

Chapter 7 


The method of performing any given task is probably the most 
important factor affecting the efficiency of the worker. We shall 
include under the heading of method any intrinsic aspect of the job 
which may be changed without changing the nature of the product. 
Thus there may be a choice of tools to be used, a choice of designs 
for any given tool, variations in the movements or posture of the 
worker, the order in which various parts of the task are carried out, 
the rate at which he works, and the load which he carries. 

The problem of choosing the particular combination of these 
factors which will be most effective has been frequently left to acci- 
dent, tradition, or unguided opinion. In recent decades, and particu- 
larly in very recent times, there has been a trend, however, toward a 
more systematic study of these questions. Motion and Time Study, 
or Methods Engineering, includes a body of rules and procedures for 
designing more adequate ways of doing a job, especially the more 
routine types of work in industry. 

Many of the questions which arise in Methods Engineering are 
strictly engineering problems. Such questions are, for example: 
*'How fast can a given machine be operated?" ^'Which kind of tool 
will cut most rapidly and effectively?" ''How can the machines in this 
plant be arranged to minimize the waste space, transportation, and 
storage of materials?" Unless the answers to these questions affect 
the performance of the worker, they are not our concern. Conse- 
quently our discussion of Methods Engineering will be concerned 
only with certain phases of the total field. 

Along with procedures for answering questions like those just 
cited, Methods Engineering contains a body of suggestions for the 
most effective use of the worker, this part of the field usually being 
known as ''motion study." In brief, motion study consists in an- 
alyzing the movements to be made by a worker in doing a certain 
task, deciding upon the best method, trying it out in order to de- 
termine whether it produces worth-while improvements in perform- 
ance, and training the worker to use the new method. In designing 




[Ch. 7 

the new method, the stated aim of the engineer is to reduce the effort 
required to produce each piece, so that production can be increased 
at no additional cost to the worker. 

Much of the early development of these ideas is credited to Gil- 
breth, who set down many suggestions which are still in use. The 
engineer first makes a detailed study of how the job is now being 
done, taking motion pictures, timing the worker with a stop watch, 
and sometimes using other devices to secure accurate measurement 
and description. This analysis commonly results in a micromotion 
chart, or process chart. The analyst then examines each phase of the 
task critically, in order to determine whether unnecessary time or ef- 
fort is spent on the phase. In making his recommendations for 
change he is guided by certain general ^'principles" which were first 
drawn up by Gilbreth and have since been modified and expanded 
slightly. The best way to realize the nature of these principles is to 
see the list in its entirety, and then to see examples of jobs to which 
the principles have been applied. 

Barnes, writer of a recent textbook in this field, summarizes the 
rules of motion economy in the following list ^ : 

(a) Rules for the Use of the Body 

1. The two hands should begin as well as complete their therbligs ^ at 
the same instant. 

2. The two hands should not be idle at the same instant except dur- 
ing rest periods. 

3. Motions of the arms should be in opposite and symmetrical direc- 
tions, instead of in the same direction, and should be made simul- 

4. Hand motions should be confined to the lowest classification with 
which it is possible to perform the work satisfactorily. 

5. Momentum should be employed to assist the worker whenever pos- 
sible, and it should be reduced to a minimum if it must be overcome by 
muscular effort. 

6. Continuous curved motions are preferable to straight-line motions 
involving sudden and sharp changes in direction. 

7. Ballistic movements are faster, easier, and more accurate than 
restricted (fixation) or "controlled" movements. 

1 Reprinted by permission from Ralph M. Barnes, Motion and Time Study, 
(2nd Ed.), 1940. New York, John Wiley & Sons. Inc., p. 144. 

2 The term therblig designates a single element of motion. 

Ch. 7] 



8. Rhythm is essential to the smooth and automatic performance of 
an operation, and the work should be arranged to permit easy and natural 
rhythm wherever possible. 

(b) Rules for Arrangement of Work to Promote the Above 

9. Definite and fixed stations should be provided for all tools and 

10. Tools, materials, and controls should be located around the work 
place and as close in front of the worker as possible. 

11. Gravity feed bins and containers should be used to deliver the 
material as close to the point of assembly or use as possible. 

12. Drop deliveries should be used wherever possible. 

13. Materials and tools should be located to permit the best sequence 
of therbligs. 

14. Provision should be made for adequate conditions for seeing. 
Good illumination is the first requirement for satisfactory visual percep- 

15. The height of the work place and the chair should preferably be 
so arranged that alternate sitting and standing at work are easily possible. 

16. A chair of the proper type and height to permit good posture 
should be provided for every worker possible. 

(c) Rules for Design of Tools and Equipment 

17. The hands should be relieved of all work that can be performed 
more advantageously by the feet or other parts of the body. 

18. Two or more tools should be combined wherever possible. 

19. Tools and materials should be pre-positioned wherever possible. 

20. Where each finger performs some specific movement, as in type- 
writing, the load should be distributed in accordance with the inherent 
capacities of the fingers. 

21. Handles, such as those used on cranks and large screw drivers, 
should be designed to permit as much of the surface of the hand to come 
in contact with the handle as possible. This is particularly true when 
considerable force is exerted in using the handle. For light assembly 
work, the screw-driver handle should be so shaped that it is smaller at 
the bottom than at the top. 

22. Levers, crossbars, and handwheels should be located in such posi- 
tions that the operator can manipulate them with the least change in 
body position and with the greatest mechanical advantage. 


All these rules can be roughly summarized in a single sentence : 
Arrange the work so that the worker can use the smallest number of 
movements and muscular contractions possible, ensuring that the 
movements which remain are symmetrical and smooth, with a mini- 
mum of fumbling and variation. More specific rules are needed, 
however, in order to remind the beginner how he is to accomplish 
these general results. The trainee also needs a considerable amount 
of observation of these methods in practice, so that he can more 
readily learn the technique of applying them. For this reason the 
mere listing of the rules as we have presented them here must be 
supplemented by films, or, better still, by practical laboratory courses 
in the practice of motion study. 

Criticisms of Motion Study. — Most of the criticisms which have 
been leveled at the Methods Engineer involve the contention that the 
methods are not more efficient, from the standpoint of total efficiency 
as we use the term in this book. Nevertheless it should be recognized 
that the aim of the Methods Engineer is to increase efficiency. 
Whether or not the aim is achieved is the main question before us 
in this chapter. 

Let us first examine how the Methods Engineer evaluates the ef- 
ficiency of a new method. The usual procedure seems to be the fol- 
lowing: The engineer takes motion pictures of the present method of 
doing the job, or he times it with a stop watch. In either case he 
records the results of only a small number of complete operations 
("cycles"). In fact, if his analysis is based upon motion pictures, 
he may obtain his time records from only one cycle which appears 
to him to be "typical." There is rarely any statistical analysis of the 
results to determine their reliability. After the Methods Engineer 
has devised his new method, a worker is trained to use it, and after 
some practice the time record is taken again, as before. 

Thus the decision about the value of the new method is based upon 
speed measurements taken under special circumstances, with a small 
number of records, for only one or for a few subjects. The lack of 
reliability data is, of course, not an intrinsic characteristic of the 
method, and the deficiency could be remedied. Also it should be noted 
that the changes in method. frequently produce such large changes in 
production that reliability analysis is not so essential. The fact that 
the analysis is based upon speed, and that it is recorded under unusual 
circumstances, is essential to the procedure as it is now used. In 
other words, there is no direct measure of the cost of work. It is as- 
sumed that the worker is putting forth equal effort during both sets 



of observations, and that variations in output must, therefore, reflect 
variation in efficiency.^ 

Adverse criticisms of the method have been based largely upon 
this point, but, unfortunately for progress, they have not produced 
any more definite evidence of the cost of work than that given by the 
Methods Engineers themselves. Thus Poff enberger * points to the 
shortened cycle of work as increasing the homogeneity of the task, the 
factor which Robinson and Bills found to be a cause of rapid ''fatigue 
decrement." This is an a priori argument without any evidence to 
show that the blocking and decrement observed by Robinson and Bills 
are, in fact, increased by the greater homogeneity of the task. Until 
this question is examined experimentally, one can argue just as easily 
for the opposite position. For one thing, the normal duration of a 
cycle in an industrial task is seldom so short as the duration involved 
in Robinson and Bills' studies. For another, Robinson and Bills 
also found a factor of conflict to be important in determining decre- 
ment. It will be noted that the effect of many of the rules of motion 
study is to reduce the conflict in the task. One never has to decide 
which hand to use, which tool to use, or make any of the other 
decisions that are likely to be involved in unstandardized methods 
of work. 

Another criticism is that a new basis of payment usually accom- 
panies the change in method of doing the work if the work is on a 
piece-rate basis. ^ The reason for this is readily seen. If the task is 
now easier than it was before, the worker should not be paid as much 
per unit of output as he formerly received. An adjustment can be 
made so that the worker produces more per unit of pay, thus making 
a saving for his employer, and at the same time the worker can earn 
a larger total pay than he did under the former system. The objec- 
tion to this is based upon the belief that the new method may permit 
faster production at the expense of additional cost to the worker (by 
affecting his health or his satisfaction). In other words, the new pay 
rates may induce the worker to exert himself more in order to main- 
tain or increase his earnings, and the additional production may not 
represent a true increase in individual efficiency. 

3 In practice, the engineer may use the method of "leveling" which is discussed 
in Chapter 10. This procedure is supposed to allow for variations in effort or other 
conditions, so that it would be possible to compare rates of work and regard them 
as indices of efficiency. This method, however, is itself open to question, since it 
depends upon the judgment of the observer and does not involve any validated 
measures of effort. 

4 A. T. Poffenberger, Principles of Applied Psychology, 1942, p. 392. 

5 E. Farmer, Time and motion study, Industrial Fatigue Research Board 
(Great Britain), 1923, Report No. 14. 


tCh. 7 

This is a very difficult criticism to meet. If the pay rate is left 
at its original level, and if the work is actually made easier, then the 
worker could reduce his effort and still earn more than before. As a 
result, his production would not reflect accurately the benefits de- 
rived from the new technique of work. To reduce the piece rates, on 
the other hand, is to assume that the saving in effort brought about 
by the new method is known. The Methods Engineer assumes that 
the time records in his trial studies do reflect the relative ease of 

Another possible way out would be to compare the rates of per- 
formance in the alternative methods under a system of time payments. 
If we could be sure that the worker would not be influenced by habits 
and traditions concerning what is a normal day's production, his ef- 
fort should be reasonably constant in the two cases. The require- 
ments of a study which would clearly indicate the relative efficiency 
of the two methods are not easy to fulfil. Each method should be fol- 
lowed over a long period of time — time enough to cancel out the 
effects of practice and random variations in effort, and time enough 
to discover any cumulative effects of fatigue. In addition, the records 
should be obtained under conditions which are as normal as possible ; 
not a few cycles while the time-study man is looking on or taking 
motion pictures. These requirements are difficult enough, but still 
more difficult is that of maintaining constant motivation of the 
worker so as to ensure that he works at a comfortable pace and that 
he does not aim at some fixed total output regardless of method. 

Obviously, such an elaborate experiment could not be carried out 
each time the method of work is redesigned for each of hundreds of 
tasks. Perhaps the only feasible solution is to carry out such studies 
for some sample cases involving a variety of tasks to which the gen- 
eral principles of motion study have been applied. Such a program 
would at least provide some factual basis for discussing the criticism 
under discussion. Meanwhile it is another criticism which is debated 
by both sides on the basis of opinion and uncontrolled observations 
alone. The debate will continue along these lines until some of the 
methods of measuring efficiency employed by the methods engineers 
have been more satisfactorily validated. 

A third important criticism of Gilbreth's methods, and one which 
is still applied to his followers, is that the procedures ignore the role 
and importance of individual differences among workers. Some 
people, for example, may find it much easier to work with the two 
hands simultaneously. A person who finds it difficult to work with 
two hands might have such a high level of dexterity with one hand 



that he could compensate for the advantage others would gain by the 
two-handed method. Balchin has given an example of a study of 
three workers.® Among them, two were most rapid when using one 
method, while the third worked better with another method. There 
is also an indication, from correlational studies of various kinds of 
dexterity, that these abilities are relatively independent of one 

It should be mentioned that this phase of the problem was not 
entirely ignored by Gilbreth. He assumed that the worker would be 
selected for the task according to his capabilities. While this assump- 
tion would be all right in principle, and much has been done toward 
improving methods of selection, the required accuracy of selection is 
impossible at the present time. To select in such a way as to over- 
come the objection to motion-study methods would require that we be 
able to choose those workers who have high aptitude for a particular 
method of doing a job, as well as choose those who are well suited for 
the general kind of work. In other words, we should have to choose 
workers who are dexterous in the types of movement involved in the 
job, and in the particular combination and co-ordination required by 
the motion study. This would carry the process of selection far be- 
yond the range of practical feasibility, even if techniques of measure- 
ment were available for carrying out such a program. 

Perhaps the answer to this objection lies in the current trends in 
the application of motion-study principles. In order to give these 
ideas widespread use, many supervisors, and even workers themselves, 
are being taught the general principles of motion economy. It is 
hoped that the supervisor or the worker himself will be able to put 
these suggestions into practice without the services of a specialist in 
motion study. If this is done, we should expect that individual dif- 
ferences between workers could be more readily taken into account. 

The criticisms we have discussed have been those which applied 
even to the more careful use of motion-study methods. Needless to 
say, there are many others which would be evident in instances where 
the techniques have been carelessly or unscrupulously used. If 
changes in method do not actually make the work any easier, and if, 
at the same time, they are made the occasion for reducing pay rates 
on the job, then serious trouble will result. It is rumored that so- 
called motion studies have sometimes been no more than an excuse 

6 N. Balchin, Movement study in packing, Journal of the National Institute of 
Industrial Psychology, 1931, 5, 274-275. Also in B. V. Moore and G. H. Hartman, 
Readings in Industrial Psychology, New York, D. Appleton-Century Co., Inc., 1931, 
pp. 298-300. 



[Ch. 7 

for reducing pay rates, and as a result the terms ''efficiency expert" 
and "efficiency engineer" have among many workers come to desig- 
nate a certain disrepute. Sound motion study is not open to this 
accusation, and those seriously engaged in this work realize how im- 
portant it is that such suspicions be allayed. The criticisms we have 
discussed are not to be taken, therefore, as an indictment of the good 
faith of motion study engineers. They merely raise the question 
whether the methods have produced as complete a solution of the 
problem of efficiency as we should have, and whether the manner of 
gauging efficiency is in serious need of supplementation. 

Special Studies of the Effectiveness of Machines and Tools 

The rules numbered 20, 21, and 22 in Barnes' list are concerned 
with the optimal design of certain tools, and are only examples of the 
general problems which come under this heading. Answers to ques- 
tions of design of tools have been sought by a variety of methods, 
ranging from simple time study to the measurement of metabolism. 

A very early problem which arose in Taylor's work with heavy 
labor was that of the choice of size of shovel and load in handling 
various materials. Taylor designed various shovels which would 
pick up the proper load of a given material (21^ pounds, according 
to his analysis). The proper load was arrived at by time study. 
That is, Taylor chose the load which would permit the highest average 
rate of performance by the worker. The same problem was attacked 
later by Wenzig, using metabolic measurements to determine optimal 
load, height of throw, and distance of throw. ^ He found that the 
optimal load varied with the height to which the load was to be 
thrown. For heights over one meter, the optimal load was 7.25 kg. 
(16 lb.), while for lower heights the optimal load was 9 kg. (20 lb.), 
the latter a close agreement with Taylor's result. 

This example shows the varied procedures which may be used in 
solving questions of design. One may choose an arrangement which 
"looks good," or which has grown up by tradition. One may take 
Taylor's, approach and perform time studies with various possible 
designs, or one may use appropriate measures of the cost of work 
and so study efficiency directly. The traditional approach is not 
likely to give the best answer. The choice between the other two de- 
pends upon the frequency with which a given tool is to be used, and 
the availability of measures of cost of work. Shovels are used by so 

7 Summarized in E. Atzler, Arbeitsphysiologie : II, Ergebnisse der Physiologic, 
1939, 41, 213-215. 

Ch. 7] 


many thousands of workers that it would seem worth while to devote 
considerable time and expense to determining the basic factors which 
make for optimal design. Wenzig's study would represent a good 
start, at least, toward such an analysis. A tool which is designed for 
a specialized task could not, of course, be studied in this exhaustive 
manner, and time study would be a more appropriate method. A few 
examples of tools which would warrant considerable study are the 
typewriter, automobile and truck controls, screw drivers, and other 
common hand tools. In addition, the general problem of location and 
position of levers and cranks would have implications for the design 
of many machines. 

It may not always be necessary to use the slow and cumbersome 
method of metabolic measurement in order to arrive at a useful an- 
swer. In some cases the metabolic measures would not provide a 
satisfactory answer because the work is too light. Nevertheless it 
would be well to improve upon time study as a method of answering 
questions of design for those tools which are in very wide use, such 
as those just mentioned. 

One device which has been applied several times is to compare the 
force which the worker can exert upon the work with various tools. 
For example, screw drivers have been compared to determine the 
shape of handle with which the worker can exert the maximum torque 
upon the screw.® In the latter case, a cylindrical handle was more 
effective than a tapered handle, and larger diameters were better up 
to the maximum tested (40 mm.). 

Another somewhat similar method has been applied to the type- 
writer. Norton compared brands of typewriter in terms of the total 
physical work required for operating the various keys and levers.® 
Among the five brands tested, the worst required about 30 per cent 
more work than the best. In this case the measurement can be made 
independently of the operator — it is a measure of the force required 
to produce a given result. In other problems, as in the case of the 
screw driver mentioned above, the experiments must be performed 
with the subject since we are interested in the way the tool fits his 

Another problem of fitting the tool to the worker also involves 
the typewriter. Since the fingers differ in relative strength, the key- 

8 B. Rubarth, Untersuchung zur Bestgestaltung von Handheften fiir Schrauben- 
zieher und ahnliche Werkzeuge, Industrielle Psychotechnik, 1928, 6, 129-142. 

9 F. H. Norton, The work required to operate several makes of typewriter, 
Transactions of the American Society of Mechanical Engineers, 1928, 50, (Man- 
agement Section) 29-36. 


[Ch. 7 

board should be so arranged that the stronger fingers do more of the 
work. Hoke, by determining the relative frequency with w^hich dif- 
ferent letters are written, was able to show that the present typewriter 
keyboard is unsatisfactory from this standpoint. As Table 14 shows, 
the little finger of the left hand has more to do than the second finger 
on the right hand. As measured by tapping tests, the fingers of the 
left hand are slower than those of the right, but on our present key- 
board the left hand writes more letters than the right. Hoke and 
others ^° have proposed reorganized keyboards which have received 
some attention, but so far the difficulties of changing keyboards have 
outweighed their possible advantages. Of course the arrangement 
should take into account not only the relative frequency of individual 
letters, but it should also arrange for common letter combinations to 
be written with an easy pattern of movements — perhaps alternately 
with the two hands, as Dvorak and his collaborators have done. 

TABLE 14* 
Load ox Fingers During Typing 

Percentage of Ideal Load 














Left hand 






* R. E. Hoke, The improvement of speed and accuracy in tj-pewriting, Johns Hopkins 
University Studies in Education, 1922, No.. 7, p. 32. 

It should be recognized, however, that these studies of force re- 
quired to operate a machine, or of force exerted by a tool, are only 
partial solutions of the problem of optimal design. It is conceivable, 
for example, that a reduction of the force required to operate a type- 
writer might not always improve its efficiency. Perhaps a certain 
amount of pressure upon the keys is better than a mere touch, as in 
the case of the electric typewriter, or perhaps the pressure required is 
a very unimportant element in the fatigue of typing. The writer is 
not a professional typist, but his own experience fails to reveal any 
particular strain of the hand and arm muscles in typing — the princi- 
pal feelings of fatigue are postural and are localized in the back. The 
final test of the design of typewriters, whether from the point of view 
of pressure, or from the point of view of keyboard arrangement, will 
not come until it is possible to get accurate measurements of produc- 

10 A. Dvorak, N. I. Merrick, W. L. Dealey, and G. C. Ford, Typcivriiing 
Behavior, New York, American Book Co., 1936. 



tion under comparable conditions for the different designs, and until 
it is possible to measure the over-all effort and fatigue involved. 

A similar argument applies to the studies of screw driver handles 
and similar devices. The shape of handle v^hich permits the worker 
to exert the maximal force would be the best handle if the screw 
driver is to be used only for the exertion of great force. In many 
uses of the screw driver, however, the maximal force is exerted for 
only an instant. By far the greater portion of the time is spent in 
rapid turning against little resistance. This has been recognized, in 
that a specially designed screw driver for light work has been 

Many such problems cannot be completely solved at present, be- 
cause, as we have seen, measures of effort which are now available 
are not applicable to the kinds of work involved. In. heavy work the 
situation is more satisfactory, and several applications of metabolic 
measures have been made. We have already mentioned the metabolic 
study of shoveling, and in an earlier chapter we discussed Atzler's 
study of pushing and pulling loads on a cart. An additional example 
is a study of the use of a wheelbarrow for transporting bricks. 

Crowden measured the metabolic cost of wheeling a barrow loaded 
with bricks under varying conditions of load, rate of walking, and 
design of barrow. Variations in the number of bricks per load 
affected not only the weight to be moved, but also the balancing 
characteristics of the barrow. A load of seventy bricks was found 
to provide more efficient performance than either a smaller or a 
heavier load if the optimal arrangement of the load was employed. 
The optimal arrangement of the load, as might be expected, was 
that in which the point of balance of the barrow was a position which 
brought the handles level with the palms of the worker in his upright 
position. Since a substantial part of the energy consumption was 
attributable to the lifting of the barrow handles when starting, it was 
possible to reduce this factor of cost by increasing the length of the 
legs. By trial, it was possible to use longer legs than those originally 
provided, without interfering with the use of the barrow. 

Evaluation of Progress in Methods-Design. — The time-study 
analysis which methods engineers use to determine the relative ef- 
ficiency of methods has contributed much of practical value in in- 
dustry. In many of the applications, the increased production is so 

11 Barnes, op. cit., 234-235. 

12 G. F. Crowden, The physiological cost of the muscular movements involved 
in barrow work, Industrial Fatigue Research Board (Great Britain), 1928, Report 
No. 50. 



[Ch. 7 

obviously a result of eliminating unnecessary muscular effort that 
there is little question of the value of the improved methods. To 
take an extreme case : If the Methods Engineer finds that a third or 
even half of the operation can be eliminated completely by a simple 
rearrangement of the work, we cannot question his statement that 
efficiency has thereby been improved. 

On the negative side, however, certain questions remain un- 
answered, and certain limitations of present techniques must be 
recognized. In the less obvious cases, the changes consist of a re- 
arrangement of the pattern of work rather than an evident elimina- 
tion of parts of the task. Here a sound evaluation of relative 
efficiency cannot be made solely upon the basis of rate of performance. 
Until measures of effort are available, there is much room for argu- 
ment on the relative merits of various methods, or of various designs 
of tools and equipment. Time study is inadequate because it relies 
upon extremely crude estimates of effort by the engineer, or upon 
the assumption that effort remains constant throughout the compari- 
son of the different methods. 

It is evident that these techniques of motion economy can be 
most readily applied to routine manual and mechanical tasks. It must 
not be assumed, however, that their application is limited to factories. 
Applications have already been made with success in hotels, laundries, 
business offices, and even in the home kitchen. A highly variable 
job, such as preparing a meal, can often be broken down into parts 
which can be partially standardized. At least those portions which 
are performed most frequently can be treated according to rules for 
arranging a work place which were listed on preceding pages. 

Where the manual operations are only incidental to the main 
task — as in studying, writing, scientific work of various kinds, or 
executive functions — the tools of the trade may be arranged in similar 
fashion. Extreme disorder may detract from time available for the 
main task in hand. On the other hand, there is danger of reducing 
the manual operations too far. It may be better for the desk worker 
to walk across the room occasionally for a book or to get something 
from a file. At best, better arrangements will contribute little to the 
main task except to prevent minor annoyances or delays. In the tasks 
for which the rules of motion study were first designed, these annoy- 
ances assume major proportions and make the difference between 
economical and uneconomical performance. 

Chapter 8 


Most of the popular questions concerning psychology in daily work 
deal with motives and incentives. Why is a certain individual going 
into a decline of productivity? What appeal will be effective in rais- 
ing the output of a certain group of factory workers, students, or 
soldiers? These are the questions which make up the general topic 
to be considered here. We might also use that popular word ''mor- 
ale," but it is doubtful if we should gain anything by doing so, since 
the term is so vague and variable in meaning. 

; Recall that in our earlier definitions we said that the terms motive | 
and incentive refer in general to factors which raise or lower the 1 
level of effort which an individual puts into a task. These factors J 
are usually distinguished from others which we called governors. : 
The latter factors affect performance also, but they do so through a 
change in the capacity, or limits, of the organism rather than through 
a change in effort. The distinction is not always easy to draw in a ! 
particular situation, and often the same factor may act both as a 
governor and as an incentive. For example, hot weather may not 
only reduce our capacity for heavy exercise, but it probably also 
reduces our desire to try. 

In spite of the difficulties of maintaining this distinction in prac- 
tice, it is (or should be) the aim of any research study of work to 
determine which effect is produced by any given factor. The effect 
upon efficiency of performance will differ, and the way the factor can 
be controlled, or used in prediction, will also differ. For example, if 
we do not do so much work in hot weather because weather is a nega- 
tive incentive, and if it is found that the temperature has no real effect 
upon capacity, then additional incentives may be just as good as an 
air-conditioning system. If temperature is actually a governor of 
performance, however, additional incentives might produce lowered 
efficiency even though they do raise the level of output. 

Incentives and Efficiency. — While an incentive affects the amount '> 
which an individual does, it does not necessarily raise efficiency at the \ 
same time that it raises productivity. In most physical or muscular j 




[Ch. 8 

tasks, it is found that the energy consumption per unit of work is 
least for a moderate pace — a brisk walk rather than a run, and so on. 
If the effort of an incentive is to increase the pace of the work be- 
yond this optimal level, therefore, it may decrease the over-all effi- 
ciency of performance. In the long run, then, it may have adverse 

We may assume that any task has its optimal level of effort. 
For this reason the problem of motivation in work is not simply 
to find those incentives which will raise output. This is only a part 
of the program. In addition, we want to find those incentives which 
will raise production without unduly decreasing efficiency. For ex- 
ample, fear of severe punishment might raise the level of output, but 
at the same time this fear would waste so much effort that output 
could not be maintained at the new level without later consequences 
in the form of ill-health, decline of production, and so on. 

There is, then, the need to consider the optimum rate of perform- 
ance which is to be maintained. The practical use of incentives does 
not always aim at increasing the rate of production, but rather at 
maintaining the rate of work at a level as near as possible to the 
optimum level. 

It may be argued that in most industrial operations the level of work 
is probably below this optimal rate. If so, the problem is usually 
that of finding incentives which will increase the rate of work. 
Studies of restriction of output would seem to indicate that this is 
the normal state of affairs. Nevertheless, upon occasion, there may 
be the danger of too much effort ; the problem then becomes that of re- 
ducing incentives to the optimal level. 

Sources of Information Upon Incentives and Motives 

Results of research on psychological problems, it goes without 
saying, are always a function of the method by which the data were 
obtained. In the field of motivation and incentive, however, this 
point cannot be overstressed. Any particular situation involves so 
many complex interrelations of motives that any analysis is abstract 
and incomplete. In our survey of available facts, therefore, we must 
keep in mind how these facts were obtained. The methods of re- 
search study must be noted, along with the facts. 

Laboratory Studies of Motivation in Work. — Studying the human 
being in the laboratory has the advantage of reducing the complica- 
tions so that it is easier to interpret the results. When we wish to 
obtain information of practical value on such a problem as motiva- 

Ch. 8] 



tion, however, the laboratory situation involves factors which may 
destroy the very elements we are interested in. If, for example, we 
wish to find what makes certain kinds of work boring, the fact that 
our subjects are taking part in an experiment may make them so 
co-operative that we are unable to produce the type of monotony 
which would occur in the working situation. 

For this reason, laboratory experiments in this field have their 
main value in preliminary exploration. The results cannot be applied 
directly to a factory situation without further investigation under 
more natural conditions. With this caution, let us consider some of 
the findings which have the most practical value. 

In classroom experiments, several of the commonly recognized 
incentives have been shown to affect speed of learning and perform- 
ance on various tests. Giving the pupil a statement of his previous 
performance, without any special effort to incite rivalry, or any 
praise, produces increased output. Praise, reproof, rivalry, and the 
presence of other workers have also been shown to be effective. In 
connection with praise and reproof, further analysis of the effects 
indicates that praise is more effective for poor performers, and re- 
proof more effective for good performers. 

None of these results are very surprising, and they add little more 
than a confirmation of common sense. Another laboratory study is 
worth mentioning in greater detail because it is less obvious, and 
because it has an important bearing upon the setting of standards 
for piece-rate or other bonus methods of payment. 

The experiment to which we refer, carried out by Mace, compared 
rates of learning with and without a standard or goal set by the ex- 
perimenter.^ When the goal was fixed throughout the experiment, 
the subjects did not progress so well as subjects who were told simply 
to ''do their best." Poorest performance of all was that of a third 
group who were told to aim at surpassing their previous week's 
record. (Figure 17.) 

We should not jump to the conclusion that setting a goal for per- 
formance is poor policy. In fact. Mace draws no such conclusion, 
but instead tries to find out why these standards were not effective. 
In the group where a fixed standard (seventy units per session) was 
given throughout the experiment, he found that the standard was in- 
fluential, but only during the time when the subjects' skill had de- 
veloped to the point where they could reasonably expect to reach the 

1 C. A. Mace, Incentives : Some experimental studies, Industrial Health Re- 
search Board (Great Britain), 1935, Report No. 72. 



[Ch. 8 

100 - 
90 - 

g 70h 


0- 50 

^ 40 



J L 





I 3 5 7 9 M 


Figure 17. Practice Curves in Computation for Different Instructions 
Group A: "Subjects instructed to work with the intention of surpassing a figure representative 

of their previous performance." (37 subjects) 
Group B: "Instructed to endeavor to attain to and surpass a constant and absolute standard, 

viz., seventy correct computations in the ten-minute period." (37 subjects) 
Group C: "Instructed simply 'to do their best.' " (14 subjects) 

(From Mace, op. cit., p. 9. Reprinted by permission of the Controller of His Britannic 
Majesty's Stationery Office.) 

In a second part of the experiment, Mace carried the problem 
further by using a moving standard or goal; that is, each day the 
subject was told that he was expected to do a certain amount. This 
amount was estimated on the basis of studies of the rate of learning 
the task, and was adjusted so that the subject had a reasonable chance 
of working at the standard rate. Here the setting of a standard 
produced better results than the instruction to "do your best," the 
instruction which had produced the most rapid learning in the first 
part of the experiment.^ (Figure 18.) 

The general conclusion to be drawn seems evident. Setting a goal 
for the performance of a worker will be most effective if it is ad- 

2 Mace found that the effect of these incentives was to make the daily outupt 
curve flatter and thus increase the total output without much increase in the 
maximum rate. If we were to use output criteria of fatigue, it would appear that 
the incentives increased efficiency as well as rate of work. 

Ch. 8] 






■z. 90 


3 80 


O 70 




Figure 18. Performance in Computation under Two Instructions 

Group A: Subjects instructed to do their best to improve. (No specific standard.) (10 

Group B: Instructed to surpass a specific standard prescribed for each day. (10 subjects) 
(From Mace, op. cit., p. 21.) 

justed to his level of skill and ability. Methods of bonus payments 
which do not take this factor into account may not provide incentives 
to higher production at all. Some other studies which will be men- 
tioned later would indicate that bonus systems may actually serve as 
deterrents to production rather than as positive incentives. 

Field Studies of Industrial Motivation. — Some investigations are 
often discussed under the heading of ''morale," which is merely an- 
other term for the degree of interest and enthusiasm which is felt 
by the greater number of workers in a particular organization. There 
are several ways in which such information can be gathered. 

1. Questionnaire surveys of attitudes toward company practices 
and policies, working conditions, and supervision. Here a set of 
carefully phrased questions is prepared, and provision is made for 
the workers to hand in their replies anonymously. Much depends 
upon the wording of the questions, and upon the way in which the 



[Ch. 8 

questionnaire is ''sold" to the workers. That is, the workers must 
be convinced that their answers are entirely confidential, and the 
possibility of distorting the replies through leading questions must 
be removed. 

2. Interview surveys. Here the problems of gaining frank and 
accurate information are similar to those of the written question- 
naire. The advantage of the interview method is in allowing for 
greater elasticity of the topics considered and the types of comments 
made. There are corresponding disadvantages, however. The inter- 
viewer's bias and poorly phrased questions may invalidate his findings 
unless he is very carefully trained. 

3. Exit interviews. The worker who is leaving because he has 
been discharged, or because he is disgruntled, is likely to be willing 
to make statements which he would not make so long as he was em- 
ployed. Allowance must be made for distortions due to the emotional 
upset which may have occurred. Even so, the information is of 
great value, and may permit correction of conditions which could not 
be discovered by the management in any other manner. Such inter- 
views bring a distorted sample, however, and tell us little of the 
reactions of the majority of workers. 

4. Analysis of turnover records. The turnover rate represents an 
over-all picture of the effects of personnel policies, supervision, and 
conditions of work, in relation to economic conditions, competition 
for labor, etc. If the rate is computed on the basis of avoidable losses 
of men, it gives an even better picture of the effects of company 
policies and practices. The formula for this computation would be : 

Avoidable separations would include dismissals for incompetence 
and infractions of rules, voluntary quitting for reasons other than 
health, marriage, etc. It would not include layoffs for reason of 
shut down in a particular department, or other separations of this 
kind. The rate thus computed reflects, therefore, the value of the 
hiring procedure, the relation of wages and conditions to those in 
other industries and plants, and the conditions of work and super- 

If the hiring procedure is adequate, and if the wages are standard 
for the area and industry, then an abnormal turnover rate for a 
given department, or for a given plant, indicates some source of dis- 
satisfaction which is worth searching for. 

Turnover rate per month 

Avoidable separations 

(Average payroll per period) 

Ch. 8] 



Further analysis by department, by supervisors, by various types 
of worker may aid in the discovery of this factor. Scott, Clothier, 
Mathewson, and Spriegel,^ for example, show an analysis of turn- 
over rates for workers of various degrees of intelligence in certain 
organizations. In one organization the highest turnover was for the 
workers of higher intelligence, and there was reason to believe that 
these were more valuable workers on the job involved. Their con- 
clusion was that the supervision had not been adapted to the needs 
of this group of employees. 

5. Naturalistic studies. Interesting hints upon industrial motiva- 
tion have been gained by men who have simply gone to work in 
various plants, gained the confidence of fellow workers, listened to 
conversation, and asked questions. Such reports cannot be con- 
sidered as the last word on the subject, since too much depends upon 
the individual bias of the observer, the chance experiences which he 
has, and other uncontrolled factors. Nevertheless the observations 
are valuable even if they are considered only as a first step to be 
followed up by more accurate methods later. 

Mathewson's study of restriction of output in industry is an ex- 
ample of the naturalistic type of study at its best.* He reports a 
series of cases in which workers were producing at a level well below 
what they might have done without overexerting themselves. The 
important factors which he regards as responsible for this restriction 
are (1) the desire of the supervisor to maintain as large a staff as 
possible in order to meet future emergencies; (2) wage "incentive" 
methods like the piece rate, plus the fear that the rates will be cut if 
the worker earns too much ; ( 3 ) the report that a time study is to be 
taken, as well as the actual presence of a time-study man; (4) fear 
of unemployment, that is, the desire to make the work last as long as 

The general impression one gets from reading Mathewson is that 
many wage payment plans have been set up by someone who is think- 
ing in terms of "The Economic Man," someone who assumes that 
workers will always do what is to their greatest immediate financial 
advantage. In other words, the payment methods and other policies 
seem to be set up without much understanding of the worker. Not 
only that, but management is often unaware that the payment method 
is unsuccessful, even after it has been put into practice. Mathewson 

3 W. D. Scott, R. C. Clothier, S. B. Mathewson, and W. R. Spriegel, Personnel 
Management (3rd Ed.), New York, McGraw-Hill Book Co., Inc., 1941, p. 502. 

* S. B. Mathewson, Restriction of Output Among Unorganised Workers, New 
York, Viking Press, Inc., 1931. 



[Ch. 8 

gives instances in which he interviewed the managers of a plant and 
found them confident that restriction of output was a minor problem 
in their plant because of the new method of payment. In some in- 
stances these were the same plants in which Mathewson had obser^-ed 
definite restrictive practices. 

Mathewson is aware of the limitations of this kind of study. He 
notes, for example, that he has no way of estimating the prevalence 
of restrictive practices. He vouches for the cases which he reports, 
but he cannot say how typical they are. For such information it is 
necessary to turn to other sources. 

The first two of these methods of field study are better adapted 
to discover facts on motivation which would have wide application. 
Methods (3) and (A) are used primarily to find specific sore spots 
in an organization, information of great importance to the organi- 
zation concerned but not of as great value to us in our search for 
general information on the important incentives for industrial 
workers. At least they have not produced so much up to the present 
time. Method (5). as we have suggested, is not a \try reliable pro- 
cedure. Therefore we shall devote the major discussion to question- 
naire and interview studies. 

General Results of Questionnaire and Interview Studies 

A number of these investigations have asked the worker to rate, 
or rank, the relative importance of various factors in the work situa- 
tion as incentives for remaining at the job or as factors which would 
determine his changing to a new job. This is a ver\' difficult question 
for anyone to answer, but it was hoped that careful phrasing of the 
question, and securing answers from large numbers of workers, 
would cancel out some of the possible errors. Whether or not this 
is true, the results are interesting as indicating what the workers 
think of as important influences, even though there may be other less 
concrete factors which are not mentioned. 

Goodwin A\'atson has summarized several of these sets of data/ 
and the following are reproduced from this source. 

Table 15, below, shows the ranks which were assigned to various 
factors in a poll taken by Fosdick. The first column reports the rela- 
tive standing of the items in the voting by a large number of em- 
ployees, while the second column reports the employers' opinions as 
to the relative importance of the factors as incentives. Thus the 

5 Good\^-in Watson, '"AVork Satisfaction," Chapter 6 in Industrial Conflict, Year- 
book of the Society for the Psychological Study of Social Issues, 1939, pp. 114-124. 

Ch. 8] 



study shows something of employers' misunderstanding of motiva- 
tion; it also gives us useful hints as to which factors the workers 
regard as important. Of course we cannot say how other factors 
might have been placed in relation to the eight mentioned, since these 
eight were the only ones listed in the question. 

TABLE 15 * 

Rank Assigned Various Factors in Morale by Employers 
AND Employees 

Employee Employer 

Morale Item Ranking Ranking 

Credit for all work done 1 7 

Interesting work 2 3 

Fair pay 3 1 

Understanding and appreciation 4 5 

Counsel on personal problems 5 8 

Promotion on merit 6 4 

Good physical working conditions 7 6 

Job security 8 2 

♦Watson, op. cit., p. 119. 

In Table 16 Watson shows the results of a study of "225 young 
men, all American born, unmarried, high-school graduates, who had 
had two or more full-time jobs for at least a year, but who were un- 
employed in 1933-34 .... They were asked, 'Which of the jobs you 
have held appealed to you most or proved most interesting?' and 
'Why?' " The reasons mentioned most frequently are given in the 
table. One hundred fifty-seven out of the 225 gave classifiable 

TABLE 16 1 

Reasons Assigned by 157 Young Men, High-School Graduates, 
FOR Preferring One Job Rather Than Another 

Reason Per Cent 

( 1 ) In line with vocational aspiration 29 

(2) Congenial contacts with people 24 

(3) Like responsibility, chance to use initiative, prestige 19 

(4) Variety 12 

(5) Opportunity for promotion 8 

(6) Salary better 4 

(7) Shorter hours 4 


t Watson, op. cit., p. 120. 



[Ch. 8 

The general pattern which emerges from these data should be kept 
in mind by every supervisor and executive. Wages turn out to be a 
secondary consideration, and we shall present other evidence on this 
point a little later. In general, the worker seems to want a certain 
degree of independence and initiative, plus recognition for his work 
and his value to the organization. He wants a supervisor who guides 
and directs rather than commands. 

The failure or disappointment involved in many ''welfare" 
schemes in industry has probably resulted from a failure to take 
these factors into account. A fancy clubhouse, or a much-publicized 
profit-sharing scheme may be unsuccessful because it neglects or even 
prevents attainment of some of the above conditions. The clubhouse 
does not make up for irritating supervision, and it may appear as 
charity rather than as recognition of the value of the worker's per- 

The greatest value of these questionnaire studies of motivation 
or attitude is probably not to be found in these very broad questions 
which try to get a composite picture of factors for a variety of jobs 
and conditions. Instead, the greatest benefits come from information 
which is so specific that it does not lend itself to a general summary 
of this kind. A particular source of friction in Department A of 
Company B may not shed much light upon the general problem of 
motivation, but it may be extremely important to the efficiency of that 
organization. Perhaps from the point of view of benefits to effi- 
ciency, the discovery of this sort of specific information is more im- 
portant than the attempt to discover generalizations and principles. 
If that is so, it is more important that the applied psychologist be 
trained in the techniques of developing and administering reliable 
questionnaires than that he have a consistent and all-embracing theory 
of the motives of the worker. 

The Western Electric Studies. — The Western Electric Company, 
in collaboration with the Harvard Business School, has been carrying 
on extensive investigations related to the topic of this chapter. In 
fact they are the most elaborate studies of the kind ever carried out; 
it would be impossible for us to do full justice to them here. Roeth- 
lisberger and Dickson have published the most recent summary of the 
results in a large volume.^ This was summarized by Stuart Chase in 
"The Reader's Digest." ' 

^ F. J. Roethlisberger and W. J. Dickson, Management and the Worker, 1939, 
Harvard University Press. Also summarized in the National Research Council 
Report, Fatigue of Workers, 1941. 

The Reader's Digest, (February) 1941. 

Ch. 8] 



In brief outline, they started to investigate the effects of such 
factors as Hghting and rest periods upon production. Their pro- 
cedure was to set up a special experimental room within the factory 
to which certain operators were transferred to do their usual work. 
After many months' work with various combinations of rest and 
work, they found that production was markedly improved. (Figure 
19.) When they returned to the original conditions with no rest 
periods, however, they found that production remained at a high 
level rather than returning to the original value. 

A search for the reasons led them to the conclusion that the im- 
portant factor in the improved production was not the rest periods. 
Instead, whatever effect the rest periods had was covered up by a 
much more complicated set of factors. The operators had been given 
special attention ; they were allowed more freedom than they had in 
the regular department ; the supervisor was more of an observer than 
the type of supervisor they were used to; and the operators them- 
selves had become a more congenial and closely knit group. 

The results led to an extensive interview study of several thousand 
workers in the plant. Special interviewers were trained to gain the 
confidence of workers and convince them that what they said would 
be kept anonymous. Any specific complaints were referred for in- 
vestigation and correction if possible. Comments about the super- 
visors were brought up for discussion in supervisory training classes 
with the identifying information concealed, so that the supervisors 
could see what effect some of their acts and policies had produced. 

In addition to the benefits produced in these ways, the investigators 
believe that another result was just as important. The workers felt 
better for having been interviewed in this manner, even if there was 
no evidence that their statements had an effect. They felt that an 
interest was taken in their feelings, and the interview gave them a 
chance to ''blow off steam" on topics which they otherwise could not 
mention and which they might brood over indefinitely. 

A substantial part of the value of the published reports we have 
mentioned cannot be reproduced in a brief space. Interesting ob- 
servations on social organization within departments are presented. 
Several of the interviews with workers are reproduced in detail, and 
reading them gives an insight into the method and the results which 
can be gained in no other way. The reader is therefore urged to 
turn directly to Roethlisberger and Dickson's report. 

Evaluation of the Western Electric Studies. — Since the "Haw- 
thorne experiments" (as these studies are often called) have received 

Exp. Periods 111 1 1 3 Ulslel 7 I 8I9I »o I 11 I it I >5 



















Operator 3 

< 0| 


oerotor A 


JCroTor , 

> » 




5 < 

i 7 

8 < 

? K> 




Exp. Periods 

Weeks Aor Jun Ati9 Oct Dec f<i> Mar May July 3epr tier* Jan Mor Apr Jun 

Tr y> '2S ZO IS K) 4 31 26 21 15 O 5 Z 27 U 
ending ^ Jon Mqt A^^ Jur> Aug oTi 0^ ^ May 

Saturdays 2» 25 i? u 7 i H u » a 9 t » 25 

Figure 19. Average Hourly Output in Western Electric Company Test Room 


(See table on following page.) 


Ch. 8] 



so much public attention, we may well consider just what their psy- 
chological importance is. In the first place it might be said that their 
most important findings were obtained because they began with a 
faulty experimental method. It is not surprising that any effects of 
rest periods were covered by more complicated motivational factors 
when we note that the subjects of the experiment knew that an ex- 
periment was going on and that rest periods were the subject of 
investigation, and when we recall that those participating in the ex- 
periment were given special privileges and attention. If we wanted 
to find the effects of rest periods per se, these would be disturbing 
factors which have to be controlled. Before the value of rest periods 
could be assessed, it would be necessary to introduce the rest periods 
with as little change from usual working conditions as possible. Also, 
the conditions would have to be maintained for a long time so that 
the effect of novelty, and the feeling that management was taking 
greater than normal interest in the worker's welfare, would wear off. 

We have said that the results were not surprising. They are not 
surprising now, but the more important question is whether they 
should have been anticipated before the experiments were started. 
We cannot answer this now, although it seems that it should at least 
have been suggested at the time. 

It may be true that management should have been aware that 
workers who feel that their work is important, who are given a chance 
to express their preferences and opinions, and who feel themselves 
free from overly strict supervision would work more effectively. 
Those who planned these experiments probably would have agreed, 

Data Pertaining to Figure 19 

Experimental Per Cent Decrease 


Experimental Conditions Working Hours 

from Standard 


in Weeks 

of Work per 

Week ♦ 

Working Hours 















Two S-min. rests 





Two 10-min. rests 





Six 5-min. rests 





15-min. a.m. rest and lunch 
10-min. P.M. rest 





Same as VII, but 4:30 stop 





Same as VII, but 4:00 stop 





Same as VII 





Same as VII, but Sat. a.m. off 









Same as VII 



* By "experimental working hours" is meant the total time lapse between official starting 
and stopping time for the day (standard working hours), from which those time decreases 
due to the experimental conditions of work listed above have been deducted. 

(Reprinted by permission of the publishers, from Roethlisberger and Dickson, op. cit., 
p. 76.) 



[Ch. 8 

in principle, to such a statement at the time. They apparently did 
not, however, realize that their experimental procedure was going to 
bring about these conditions. 

Perhaps more important than any actual contributions to our 
fundamental understanding of industrial motivation is the fact that 
the research indicates a trend on the part of management to become 
interested in these problems and to do something about investigating 
them. In addition, the wide interest which these studies has aroused 
may pave the way for further and even more enlightening studies. 

The Western Electric Company itself felt that it had benefited 
in its supervisory relations as a result of the test room studies and 
the elaborate interviewing programs which grew out of them. What 
had been an abstract statement became more concrete and intimately 
related to their work and experience. When we try to draw general 
statements or principles from these studies, however, we go back to 
very general statements which seem self-evident. 

Perhaps industrial psychological research has other values for the 
organization which carries it on, beyond the factual and generally 
applicable information which is obtained. Certainly it studies a situ- 
ation which may not be very closely duplicated anywhere else, and 
much less extrapolation is necessary. In addition, it may help to 
create a greater interest in psychological factors in work among the 
staff of the organization which carries on the research. 

Summary of the Main Industrial Incentives. — In order to bring 
some order into our collection of information upon incentives in 
work, it will be well to list some main headings under which these 
facts may be grouped. The following list appears to include most of 
the important factors : 

1. Pay. 

2. Standards and goals. 

3. Intrinsic interest in the task. 

4. The social setting of the job. 

(a) Supervisory relations. 

(b) Relations to other workers (competition, conforming to cus- 
toms, etc.). 

(c) Attitudes toward organization or management as a whole. 

We have some information which fits under each of these head- 
ings, although much more is to be desired. In some of these areas, 
little more than a start has now been made upon a research approach 
to the problems. In some cases a single research has been concerned 

Ch. 8] 



with a mixture of factors falling under different headings. In such 
cases it is not always possible to separate our discussion of the various 
headings, and there will be much overlapping. So far as we can, 
however, we shall summarize the available results in this fashion. 

1. Pay. No one doubts the importance of pay as a determining 
factor in the choice of jobs, or productivity in a given job. There 
has, however, been a tendency to overstress this factor in relation to 
others. The studies we have already discussed indicate that this fac- 
tor cannot substitute for many other incentives which are necessary 
to effective work. 

'Tncentive pay" systems have sometimes tended to overemphasize 
the importance of pay, and in addition they have assumed that the 
worker will be influenced by immediate returns. Studies of restric- 
tions of output indicate that where pay is an important factor, the 
expectation of long-term average earnings is more important than any 
immediate return which might be temporarily gained by extra effort 
at the moment. This is one way of summarizing Mathewson's find- 
ing that workers are unwilling to work near capacity if it means high 
earnings for a time, followed by a reduction in rate. In addition, the 
worker is unwilling to earn more at the expense of disapproval of 
other workers. 

2. Standards and Goals. Mace's study indicates the import- 
ance of setting standards for the work which are adjusted to the 
present ability of the worker. Our discussion in Chapter 10 of the 
methods of time study now in use in industry for determining stand- 
ards of performance, as well as Mathewson's study of restriction of 
output, indicate that this aim is by no means reached by current 
industrial methods. 

3. Intrinsic Interest in the Task. This factor is usually dis- 
cussed in industry in terms of the opposite of interest — monotony, 
dislike of the work, dissatisfaction with the nature of the work. 
Primary determiners of interest or lack of interest appear to be mat- 
ters of individual differences. A number of researches have been 
concerned with the discovery of the individual characteristics which 
make for more satisfactory adjustment to work on a routine, repeti- 
tive task, and these will be discussed in the next chapter. 

Aside from these matters of individual variation in interest and 
boredom, a few researches have analyzed factors which decrease 
boredom for most individuals — general factors of the kind which 
are the main concern of the first portion of this book. 


[Ch. 8 

Thus it has been found that performance on routine tasks can be 
increased, and the reports of the workers upon their own boredom 
improved, by certain revisions of the working procedure. Piece-rate 
payments seem preferable to time payments from this point of view. 
It is reported better to supply the materials in lots, and perhaps to 
allow the workers to stop work in order to secure a new supply, rather 
than keep them continuously supplied with material. In some cases 
it is possible to allow workers to alternate jobs. This is helpful if 
the time spent on each job is adjusted to the proper period. If it is 
either too long or too short, no benefits may be gained. It appears 
that the change of occupation need not be very great. Perhaps it 
helps to change the product slightly, even though the general nature of 
the work is the same. 

The improvements resulting from changes of the sort just de- 
scribed are modest at best, but they are probably worth trial for some 
of the most repetitious tasks. 

The dissatisfaction due to the nature of the work is difficult to 
separate from that due to a variety of other aspects of the working 
situation, such as the supervision, company labor policies, and the 
like. Not only that, but the attitude of interest toward the work is 
a function of these other phases. Not only may the worker dislike 
the supervisor, but he may also dislike the work he is doing because 
of his relations to the supervisor rather than because of the nature of 
the task itself. The improved relations to supervision found in the 
Western Electric Company test room studies can be regarded as an 
instance of the opposite effect. In many cases it is possible that the 
nature and arrangement of the work itself are relatively minor factors 
in determining contentment and satisfaction. The worker w^hose task 
is simple and not very taxing, and whose specific ambitions do not 
go much beyond his present status, may find that the work is neutral 
in interest. Much more of his attention is occupied with those factors 
which we have grouped together under the next topic, the social fac- 
tors in the work setting. 

Our knowledge of the psychology of interest is extremely limited 
at the present time. We have some research directed at the problems 
of monotony and dissatisfaction, as already mentioned and as ampli- 
fied in the next chapter. We also have a considerable body of in- 
formation about the expressed likes and dislikes of an individual as 
predictive of his vocational adjustment, and something about the 
stability of patterns of specific likes and dislikes. We still have far 
to go, however, in discovering the fundamental nature of interest, 
its place in working activities, its effect upon efficiency, and the 

Ch. 8] 


factors which modify and affect it. So far, we do not even have 
techniques for studying these problems, or formulations which would 
make it clear what our research aims should be. Most of our psycho- 
logical texts which touch upon the topic contain a common-sense dis- 
cussion which might almost as well have been written by a layman 
as by a trained psychologist. 

While we know little of the nature and conditions of interest, 
there can be little doubt of its importance in efficiency. Common 
observation indicates that work of an interesting character is not 
only more rewarding, but also less costly to the individual. It may 
be that greater improvements in efficiency can be obtained through 
increase in interest than in any other manner. There is even a sug- 
gestion in one of Poffenberger's studies to indicate that the energy 
cost of work is reduced by interest.^ Measuring metabolic rate during 
mental calculation, the investigators concluded that the output in- 
creased and energy requirements diminished when the subject be- 
came absorbed in the task as an end in itself. When a monetary 
bonus was the primary goal, output increased, but the energy con- 
sumption also increased. As we have already pointed out, energy 
consumption is not an adequate indicator of efficiency in this kind 
of work because the changes are small and subject to the influence 
of many other factors. Nevertheless Poffenberger's conclusions 
furnish us with a statement of a problem which is worth much in- 
tensive investigation. 

In industry, special bonus methods of payment are more widely 
applied to those tasks which are regarded as intrinsically boring and 
distasteful. These bonuses, which are supposed to substitute for 
interest in the task itself, may reduce efficiency even if they are suc- 
cessful in raising output. Another hint of this same relationship is 
to be found in Wyatt's experimental study of different methods of 
payment.^ An experimental group of girls worked in rotation upon 
five different tasks, spending one day per week upon each of the five. 
Three systems of payment — time-rate, bonus, and piece-rate — were 
employed at different times. The effects of these incentives varied 
from worker to worker, and also from job to job. In one task, for 
example, the output under piece rate was nearly three times that 
under a time rate, while another task was increased by only one half. 
(Figure 20.) The order of effectiveness of the wage incentive for 

8 From an unpublished study by A. T. Poffenberger and G. H. Rounds described 
in A. T. Poffenberger's Principles of Applied Psychology, 1942, pp. 410-411. 

9 S. Wyatt, Incentives in repetitive work, Industrial Health Research Board 
(Great Britain) 1934, Report No. 69. 



[Ch. 8 

different jobs was almost the same as the order of preference ex- 
pressed by the workers in rating the jobs. 

Although this experiment of Wyatt's represents only a beginning 
in research along this line, it indicates again the possibility that piece 
rates or other bonuses may be applied to the kinds of work in which 
they are least effective because of the low level of interest involved 



^ 250 



I 200 







V Weighing & 
n Wrapping 

/ 'J'-^f-^^- -V 

' Packing 


Ji; \ A ✓ ^ ' Weighing 

^--i-i-. .* •* 

''•'Vf 'K' : : Unwrapping 

y •* *** 

I I I I 

123 4 5 


15 20 




Figure 20. Effects of Incentives upon Different Kinds of Work 

(From Wyatt, op. cit., 1934, p. 24. Reprinted by permission of the Controller of His 
Britannic Majesty's Stationery Office.) 

in the job. That is, we are suggesting the possibility that extrinsic 
incentives like pay, in the absence of the interest factor, increase 
output at the expense of increased cost of work, and also that they 
are not even very effective in increasing output. 

4. The Social Setting of the Job. While the importance of 
supervisory relations, interchange with other workers, and the at- 
titudes of the worker toward management are factors of recognized 
importance in effective performance of a job, experimental investi- 
gations like those of the Western Electric Company have done little 

Ch. 8] 



more than show how important these factors can be. If we look for 
specific recommendations on how these factors can be controlled and 
made the most of in practical affairs, we find little more than a state- 
ment that we should be attentive to these phases of the working 
situation, and that we should use good judgment in dealing with 
problems of this kind. 

Although the recommendations for practice are broad and vague, 
we have made some progress in developing tools for fact-finding in 
this area of motivation. By the use of interview techniques, ques- 
tionnaires, or others of the techniques discussed in this chapter, it is 
possible to get a more accurate estimate of the workers' attitudes 
toward company policies, adequacy of supervisory methods, specific 
sources of irritation and discontent in the organization, and physical 
layout of the plant. With this information available to supervisors 
and policymakers they can use their judgment and common sense on 
the basis of facts rather than on the basis of inaccurate guesses and 
hunches about worker reactions to policies. 

Because of the vast number of factors which enter into these 
social relationships of the worker, it is doubtful if recommendations 
for practice can ever include specific supervisory procedures or mana- 
gerial policies, and at the same time be generally applicable. If a 
given supervisor or manager can, however, be provided with reason- 
ably accurate and representative data upon the specific things which 
are causing discontent in the organization, he is already far on the 
road to a solution. In other words, it is likely that in the past errors 
have been due as much to inaccurate information about the current 
situation as they were to faulty notions about what to do about the 
facts that were available. The manager who is surprised because he 
has labor trouble in spite of the fact that he has provided elaborate 
''benefits" for his workers — bowling alleys, clubrooms, health insur- 
ance, etc. — is surprised because he has no knowledge of how these 
benefits are regarded by the majority of the workers, and because of 
other factors which are considered more important by the workers 
themselves. It is not that he has violated some general principle of 
motivation because these same benefits may be very effective in some 
other organization where the conditions are somewhat different. 

Chapter 9 


It is commonly recognized that the tiredness of a worker depends 
in a very intimate way upon ''emotional" and "motivational" factors, 
and that these factors are frequently more important than the effect 
of the work itself. The man who is continually irritated by his 
working mate or his superior, who is in fear of discharge or of 
serious accident, who feels that he is doing unimportant work which 
stamps him as a failure, or who actively dislikes the tasks he is re- 
quired to perform — such a man is likely to be exceedingly weary at 
the end of the working day, and indeed may feel ''worn out" most 
of the time. 

In view of the great importance of factors like those we have just 
described, coupled with the difficulties of measuring effort and fatigue 
which we have previously discussed, there has been a tendency on the 
part of some writers to pay little attention to the nature of the work 
and its effects upon the organism, unless the work requires severe 
muscular exertions or is performed under extreme conditions of 
heat, pressure, oxygen tension, or the like. The National Research 
Council Committee which published "Fatigue of Workers," ^ for ex- 
ample, devoted the greatest portion of that report to a discussion of 
the Western Electric Company studies which are described in the 
preceding chapter. 

Because of this trend to emphasize the "attitudinal" and "emo- 
tional" side of the working situation, we must attempt to relate these 
factors to the general discussion of efficiency and the cost of work 
previously presented. It will be necessary to proceed on a common- 
sense basis without adequate support from factual studies because 
that is the present status of this problem. 

/"Nervousness," emotive predicaments, boredom, and distaste for 
the work contribute to decreased efficiency in a multiple fashion. 
First, the useful accomplishment of the worker in terms of "objec- 
tive" output is partly or wholly canceled by the dissatisfaction which 

1 Committee on Work in Industry, National Research Council, Fatigue of 
Workers: Its Relation to Industrial Production, New York, 1941. 


Ch. 9] 



he feels. In other words, the net value of the accomplishment is low 
if we count as output not only the pieces produced but also the satis- 
faction which the worker derives from that production. Second, 
effort is reduced, and consequently the objective output is also 
diminished. Third, while the effort directed toward the work is 
reduced, the emotion involved in the day's activities costs the organ- 
ism a great deal. At least the feelings of tiredness are markedly 
affected in our common experience, and we should expect that these 
feelings must reflect some change in total bodily economy. \ 

While a common-sense analysis of the influence of these factors 
upon eflficiency is easy to carry out, a more scientific study is extremely 
difficult. In fact, scientific study of the problem has been almost non- 
existent. As we have seen, we have the beginnings of an analysis of 
the factors which produce dissatisfaction and irritation in industrial 
work. One does not have to look far to find very potent sources of 
irritation. What we do not know is the effect of this irritation upon 
the organism. At this point it may be well to consider the reasons for 
this serious gap in our knowledge. There are several important diffi- 
culties involved. 

The degree of upset or irritation, for example, must be assessed by 
questioning the workers. The unreliability of such a procedure is 
well known. As a result, many studies which have touched upon the 
problem have resorted to analyses of production rather than of the 
irritation itself. In the Western Electric Company studies they drew 
their conclusions about the improved social setting of the job partly 
from casual comments and remarks of the workers, but primarily 
from the fact of continuously increasing output which could not be 
accounted for in terms of greater ease of the work. Though we do 
not doubt the general validity of the conclusions of this study, it 
should be noted that it is based upon rather unreliable criteria. As it 
happened, the experimental conditions produced such marked changes 
in the feeling tone of the workers that there was little danger of 
error due to the inaccuracy of the methods. In other problems deal- 
ing with less drastic changes in the working conditions this unrelia- 
bility becomes a serious difficulty. 

Besides the difficulties of determining when irritation and dissatis- 
faction occur in a given working situation, and also how much, there is 
the further problem of determining their effect upon the individual. 
f Aside from the low production rate, for example, and the fact that 
^he employer may lose a trained worker, does it do any harm if that 
worker is chronically bored with his task ?^ Production rate and turn- 
over are, of course, important in themselves. It is also possible that 



[Ch. 9 

chronic boredom which cannot be escaped may lead to deterioration 
in the worker. The effects, however, are Hkely to be extremely com- 
plex; they cannot be discovered by physiological measurements of the 
worker before and after a day's work. There are likely to be pro- 
gressive changes in his general adjustment at home and in his hours 
of recreation. Alcoholism, divorce, or petty fights and disputes may 
have some basis in the worker's dissatisfaction with his job. Occu- 
pational neuroses may develop.^ While these relationships appear 
obvious, nevertheless it is still impossible to determine the exact 
nature and frequency of the dependence. 

I We shall presently examine the researches which have been de- 
voted to the problem of boredom in work. \ Before doing so we must 
discuss the place of studies of fatigue and effort in relation to these 

First let us recall that the cost of work is conceived in this book 
as including all adverse effects of the work upon the worker. Thus, 
at least in principle, we have provided a place for ''nervous" fatigue 
and monotony in our general formula for efficiency. Moreover, some 
of the measures of cost of work which have been described and 
analyzed make no distinction as to the origin of the effect — whether 
in the working activity itself or in its surrounding conditions. Cir- 
culatory changes such as pulse rate and blood pressure, and changes 
in muscular tension will obviously be a function of emotive factors 
in the work as well as of the effort required by the task. In the labora- 
tory studies in which these measures have been used, the social and 
emotional factors have been held at a fairly constant level. The pat- 
tern of changes which comes about in work with disagreeable ac- 
companiments in the surrounding situation will require investigation. 

We are handicapped in our analysis of this problem by the fact 
that most experimental work upon the bodily changes in emotion has 
concerned temporary emotional upsets rather than the long-drawn- 
out but milder anxieties, fears, and angers frequently involved in the 
normal working situation. Our interest is in the repetition of emo- 
tional seizures, or the chronic emotional strain due to conflicts with 
the supervisor, distasteful work, resentment against management 
policies, and the like. Whether this kind of upset has organic effects 
similar to those which occur in the sudden startle or fear of the 
laboratory situation is difficult to determine. It is quite likely that 
there are permanent functional-organic changes — circulatory, vis- 

2 See M. Smith, M. Culpin, and E. Farmer, A study of telegrapher's cramp, 
Industrial Health Research Board (Great Britain), 1927, Report No. 43, and 
M. Smith and M. Culpin, The nervous temperament, ibid., 1930, Report No. 61. 



ceral, and glandular — which result from these chronic emotional up- 
sets, but so far we have little experimental evidence to indicate their 
nature. The reason for our lack is obvious. We cannot set up such 
situations on a deliberate experimental basis. To follow the effects 
of the natural working situation involves so many complex and un- 
controlled variables that it is unmanageable. 

Two lines of indirect evidence may eventually lead to techniques 
applicable to the working situation. Physicians in the field of psycho- 
somatic medicine have recently aroused an interest in the emotional 
factors contributing to organic disease (asthma, heart disfunction, 
digestive disturbances, and so on).^ The disturbing factors most 
frequently cited involve the family, marital, and general social re- 
lationships of the individual rather than his job situation. Never- 
theless it may later be possible to assess the importance of emotional 
factors in work as contributing to psychosomatic disorders. 

The other line of indirect evidence comes from studies of animal 
behavior. It is possible to develop ''experimental neuroses" in lower 
animals, and to study their visceral and circulatory functions both in 
the experimental situation and outside.* While these studies are 
making important strides toward the achievement of a better under- 
standing of many aspects of neurosis, it is too early to attempt the 
dangerous step of applying the results to the complex conditions of 
shop and office.^ 

The gross statistical indices of cost of work in industry also rep- 
resent composite effects of all the conditions of work. In fact, turn- 
over rates and absenteeism in industry may often reflect the emotional 
factors more clearly than they do the effort required by the task. 
Similar relationships probably hold for health records, although the 
factors in this case are more complex. These gross measures cannot, 
of course, reflect individual reactions to the working situation. It is 
only when a substantial proportion of the workers are affected by 

3 H. F. Dunbar, Emotions and Bodily Changes (2nd Ed.), New York, Columbia 
University Press, 1938. 

* H. S. Liddell, Conditioned Reflex Method and Experimental Neurosis, Chapter 
12 in Personality and the Behavior Disorders, J. McV. Hunt, Editor, New York, 
The Ronald Press Co., 1944, pp. 389-412. 

5 There are evident similarities between some of the situations which produce 
animal neuroses and those which occur in jobs where the individual must remain 
at work performing a task which he dislikes or cannot perform adequately. There 
are probably similarities in the effects of such situations in lower animals and in 
man, but we do not know how close the similarity may be. Liddell, for example, 
has experimentally produced in goats a "stiff leg" reminiscent of telegrapher's or 
writer's cramp, as well as conditions of "anxiety." The factors which produce 
these conditions in animals are certainly not identical with those which produce 
similar symptoms in man. 



[Ch. 9 

poor supervision, unfair management policies, or distasteful tasks 
that these conditions are reflected in turnover and absentee rates. 
Variation from individual to individual in the effects of these factors 
must be studied by more refined methods. 

In the Western Electric Company studies there is no direct evi- 
dence on the cost of work in the test-room experiments. Inadequate 
measures of fatigue were tried out with negative results (output 
curves for the hours of the day and the days of the week, vascular 
skin tests, blood pressure tests, and speed tests on the job). From 
these results they conclude: ''There is no evidence in support of the 
hypothesis that the increased output rate of all these operators during 
the first thirteen experimental periods was due to relief from fatigue. 
This hypothesis, which at first seemed most plausible, had to be 
abandoned." ® 

We may accept the conclusion that the primary reason for the 
increases in output in the Hawthorne studies was the improved in- 
centive situation. That is, the workers continued to exert more and 
more effort in their jobs. In addition, it is possible that the improved 
incentives led to further increase in skill in the tasks, so that not all 
of the increased output must be attributed to increased effort. In 
other words, toward the end of the experiment the workers' capacities 
may have increased so that they would have produced more even at 
their original level of effort than they had at the beginning of the 
experiment. This latter possibility does not appear to have received 
much attention in the Western Electric Company's interpretation of 
the results. 

If the workers were exerting more effort under the experimental 
conditions which they liked and which made them feel important, it 
does not necessarily mean that we should ignore the cost of work 
altogether. The workers' output improved, but, as we have already 
pointed out, this may either raise or lower efficiency. The only basis 
which the investigators had for implying that efficiency was greater 
in the test-room studies, was qualitative reports. The workers 
seemed to enjoy their work more. The physical examinations did 
not reveal any physiological harm resulting from the work, although 
we should not expect them to show anything unless the work were 
producing a really drastic physiological upset. To be sure, the 
workers were told to "work as they felt," and therefore they sup- 
posedly felt more like working, as the experiment progressed. It could 
therefore be argued that if the cost of the work had been great they 

^ Roethlisberger and Dickson, op. cit., p. 127. 

Ch. 9] 



would not have reacted in this manner. The detailed reports of the 
experiments, however, throw some doubt upon the degree to which 
these instructions were adopted. Two of the original workers were 
dropped from the group because they insisted on ''working as they 
felt," and so held down the group's monetary bonus. 

Thus all the information upon effort and cost of work in the 
Western Electric studies is vague, qualitative, and inconclusive. This 
conclusion is not to be taken as a specific criticism of the researchers 
who were responsible for these studies. It is symptomatic of the state 
of the field that this study has attracted so much favorable comment, 
not only among industrial administrators but among psycholo- 
gists and physiologists as well. The implication is, therefore, that 
we are not yet ready to carry out anything more than crude and 
qualitative studies of the "nervous" costs of work. The danger in 
the Hawthorne studies lies only in the tendency of some readers to 
fail to recognize the existence of the more fundamental problem of 
the effect of motivational and attitudinal factors upon the efficiency 
of the worker. 

A more fundamental understanding of "nervous fatigue" will 
probably come from systematic studies of autonomic function. The 
crude measures of output, turnover, absenteeism, and worker atti- 
tudes can be considered only as rough and ready devices which may 
aid in practical personnel work. To characterize "nervous fatigue" 
and to state the directions our search for understanding should take 
is, however, extremely difficult. The symptoms of headache, indi- 
gestion, lassitude, insomnia, tremor, and irritability, all suggest that 
the autonomic changes which characteristically accompany emotion 
leave residues in the same functional systems after ,the work and the 
predicament are over. It is possible that the autonomic systems them- 
selves become fatigued with prolonged emotional strain, although we 
do not know the mechanism. The problem is complicated by the fact 
that we are dealing with systems which are continually active, and 
which do not require "rest" in the sense that the skeletal musculature 
requires it. The changes are probably more subtle than the changes 
in muscular fatigue, although the latter are complex enough, as we 
have seen. 

Emotional Factors and the "Work-Itself" 

In the preceding section we have criticized the tendency to over- 
look the problem of determining the cost to the individual of emo- 
tional strain and dissatisfaction. While it has been repeatedly stressed 



[Ch. 9 

in current personnel literature that these emotional factors are im- 
portant, nevertheless these problems have not been integrated with 
the concept of efficiency. It is true that v^e are not now in position 
to work out the interrelations between emotional factors and the cost 
of work, but we do feel that there is a great need to discover these 
relationships. The tendency has been, however, to avoid even a 
statement of the problem. 

We now examine another mistaken tendency which is closely re- 
lated to the one just discussed. This is the assumption that because 
emotional and attitudinal factors are so much more important than 
the cost of the ''work-itself," the latter can be considered a negligible 
factor in most industrial tasks. More concretely, it is admitted that 
the muscular strain and abnormal physical conditions in such an oc- 
cupation as that of the deep-sea diver, sand hog, or stoker cost the 
worker a great deal, even if he works under good motivation, does 
not bicker with the foreman or fight with his fellow workers, feels 
that his work is important and that he is well paid. When light 
manual tasks, clerical occupations, supervisory or executive positions 
are under consideration, however, it is implied that there is no fatigue 
or other element of cost to be attributed to the performance of the 
task unless the worker does not like the work, has a feud with his 
foreman, or becomes involved in some other social and emotional 

When we speak of the cost of the 'Vork-itself" we refer to those 
aspects of cost which correlate with the method of doing the task, or 
with the surrounding physical conditions which are a natural part of 
the working situation. While we cannot eliminate emotional factors 
altogether, nevertheless they can be held reasonably constant in the 
laboratory and even in the factory, while we investigate the effects 
of methods of work, temperature, lighting, and the like. What we 
find objectionable in some current personnel literature is the impli- 
cation that studies of the effects of work methods or the like are of 
very little value in comparison to studies of the attitudes of workers 
toward management and supervision. 

The question to be considered boils down to this : What is the 
value of knowledge of the effect of the work itself in studying the effi- 
ciency of the worker under normal circumstances? We believe that 
this knowledge should have greater importance than is attributed to 
it in current discussions, particularly those which are based upon the 
Western Electric Company findings. 

Perhaps our point can be made most clearly by an example of a 
conference which the writer once attended. The research under dis- 



cussion had to do with the physical design of an article of household 
equipment — the optimal height for the work. The researcher had 
conducted experiments upon the metabolism, pulse rate, and amount 
of shifting of body weight of a group of subjects as they worked at 
each of the experimental heights of the article of equipment. Among 
those discussing the results were several who had been impressed by 
the importance of emotion in determining efficiency; these continu- 
ally brought up the fact that the housewife's quarrels with her hus- 
band, or her dislike of household tasks, might have more effect upon 
her efficiency than the height of the equipment. While such a state- 
ment is doubtless true, it is completely irrelevant to the question at 
hand, except as a source of variation which should be controlled by 
proper experimental precautions. The number of quarrels between 
husband and wife is not going to vary with the height of the equip- 
ment at which the housewife works, and we need not consider it in 
evaluating the design of equipment. 

Of course the situation is not always quite so simple as that cited 
above. In introducing rest periods, for example, we must consider 
their effectiveness from the point of view of attitudes as well as work 
methods. Any rest periods at all may be beneficial if they reduce the 
worker's resentment toward management or his job. There are fre- 
quent cases on record in which this appears to be part of the explana- 
tion of the benefits derived from rest periods. Nevertheless the 
problem of what kind of program of rest and work to adopt must 
take into account the fatigue developed by the work itself. Rest 
periods and their optimal arrangement would be an important prob- 
lem even under ideal conditions of worker motivation. When they 
serve a double purpose of providing increased incentive as well as re- 
lieving fatigue, they should still be designed to produce the utmost 
relief from fatigue. 

In some instances there may even be a conflict between the effects 
of a factor upon the cost of the work itself and its effects upon the 
attitudes of the worker. A new method of doing a job may be re- 
sented by the workers even if it is known to make the task easier. 
Under these conditions the net outcome would be difficult to predict, 
but such a situation should not arise. If the method can be shown 
to be beneficial to the worker, it is the task of the management to 
''sell" the method in such manner that the resentment does not arise. 

In summary, it does not detract from the importance of attitu- 
dinal, motivational, or emotional factors in work if we insist that 
the cost of the "work-itself" must also be studied and that we should 
search for methods of reducing that cost. Such studies should con- 



[Ch. 9 

tribute to improved attitudes of the worker, as well as to the reduction 
of the direct cost of work. The tendency to devote exclusive atten- 
tion to the attitudinal and motivational side is likely to give the 
reader of current publications the notion that work costs the worker 
nothing of any importance. It is implied that if we can only make 
him feel well disposed toward management, and make him like his 
work, the worker can turn out vast amounts of goods for indefinite 
periods without any harm to himself. Such an implication is cer- 
tainly to be avoided. While the attitudinal side is important it is by 
no means the whole story. 

Boredom and Dissatisfaction in Repetitive Work 

We have already referred, in the preceding chapter, to the factors 
which affect a worker's interest in his work. As we indicated there, 
the principal aspect of this problem which has been studied is boredom 
and dissatisfaction in simple routine or repetitious tasks. The reason 
for concentrating on this kind of work is apparently that the more 
extreme instances of dissatisfaction are to be found in these occupa- 
tions. In fact, some have assumed that repetitious work is naturally 
boring, and have decried the tendencies of modern industry to in- 
crease the number of jobs of this kind and to reduce the amount of 
individual initiative which the worker can exercise. 

I Repetition is, of course, one of the factors which make for mo- 
notony. We must not, however, assume that it is the only one, or 
even that it is the most important factor. As we have suggested be- 
fore, it is often difficult to separate the effects of the nature of the 
work from such other aspects as the prestige value, pay, or hours of 
work. As it happens, these other sources of dissatisfaction often 
appear along with repetitiousness and so enhance its effect upon the 

! Another reason why we cannot consider that repetition is the 
primary cause of monotony is the wide individual variation in atti- 
tude of different workers toward the same job. Almost any occupa- 
tion can be boring for some persons and interesting for others. Of 
course the average level of interest varies from job to job, indicating 
the effect of factors in the task itself, but wide individual differences 
are superposed upon these job factors in determining the degree of 
satisfaction of a given worker. 

, The general problem is therefore twofold — to discover the job (or 
situational) factors which tend to reduce the general level of interest 
of the workers in that occupation, and to determine the factors which 


make an individual especially susceptible to boredom in a given type 
of work. In our general discussion of motivation, we have sketched 
briefly the factors which have been analyzed in studying the first of 
these problems — the job or situational' factors. We now undertake 
a more complete analysis of the topic of boredom, with the principal 
interest focused upon the individual factors which are very important, 
and which have received more attention from researchers. First it 
will be necessary to analyze the general problem a little more fully. 

The Nature of Boredom and Its Effects — Our problem has its 
origins in common sense, where a distinction is made between fatigue 
and boredom (monotony). It is possible, of course, that the phe- 
nomena are not distinct, and some writers in this field have come 
to the decision that we cannot distinguish between boredom and fa- 
tigue. The worker, however, is able to maintain the distinction with- 
out difficulty, and the factors which lead him to report tiredness do 
not appear to be the same as the factors which result in boredom. It 
therefore seems better to maintain the distinction until we have def- 
inite evidence that separation of fatigue and boredom has no useful 
purpose. Those who refuse to recognize the distinction are those who 
insist upon an ''operational" definition of fatigue in terms of reduced 
output. Since boredom also leads to reduced output, it cannot be 
clearly distinguished from fatigue under this definition. Our dis- 
tinction is based upon the distinction between motivational and gov- 
erning factors in performance. Fatigue reduces the capacity for 
performance, while boredom reduces the effort level — it is a reduced 
level of motivation of the worker, involving a distaste for the work, a 
desire to cease work. 

There have been a number of attempts to find the correlates of 
boredom in the behavior and productive activity of the worker and 
some studies of internal physiological changes which accompany the 
report of boredom. If these correlates could be established, they 
would be important not only in our general understanding of the 
nature of the phenomenon but might also provide us with indices of 
boredom which are more readily measurable and more reliable than 
the statement of the worker. 

The relation of boredom to the shape of the daily output curve has 
been the subject of several investigations of the British Industrial 
Health Research Board. At first they concluded that the output 
curve for a bored worker was U-shaped for each work-spell, in con- 
trast to the "typical" output curve for fatiguing and interesting work 
where the high point of production tends to be at the first quarter 


or halfway point of the spell/ In other words, the ''monotony curve" 
was pictured as being the inverse of the "typical" output curve. The 
same investigators later decided that there were at least four kinds of 
output curve representing different combinations of fatigue and mo- 
notony.^ In the extreme case, output rises progressively throughout 
the working period. (See Figure 21, d.) Here it would be concluded 
that fatigue is at a minimum, and, if practice is ruled out, the im- 
provement would be due to the improved feelings of the worker as 
quitting-time approaches. The U-shaped curve would represent a 
more mixed, but probably a more usual, kind of curve for work that 
is boring and somewhat fatiguing. (See Figure 21, b, c.) 


Figure 21. Hypothetical Work Curves 

(A) Produced when worker is interested, but suffers from "progressive fatigue." 

(B) and (C) Mixed curves. (Both fatigue and boredom.) 

(D) "Occurs when the work is distasteful and the operative is severely bored." 

(From Wyatt, Langdon, and Stock, op. cit., 1938, p. 29. Reproduced by permission of 
the Controller of His Britannic Majesty's Stationery Office.) 

In order to test this classification of work curves, Wyatt and 
Fraser classified sixty-eight workers according to the type of work 
curve which they typically showed. The degree of boredom with 
the task was assessed for each worker from a questionnaire, and 
each was rated on an arbitrary scale of degree of boredom. The 
average boredom rating is shown in Table 17 for each curve type. 
The significance of the differences in rating was not computed, but 
the authors state that the classification of the worker by curve type 
was correct in fifty-one of the sixty-eight cases. This is substantial 
agreement, when we consider that the classification was made into 
four types, and that a chance result would show about 25 per cent 
correct instead of the 75 per cent actually obtained. Such a result 
would be statistically significant. 

^(^^ S. Wyatt and J. N. Langdon, assisted by F. G. L. Stock, The effects of 
monotony in work, Industrial Health Research Board (Great Britain), 1929, 
Report No. 56. 

8 By the same authors, Fatigue and boredom in repetitive work, Industrial 
Health Research Board (Great Britain), 1938, Report No. 77. 

Ch. 9] 



TABLE 17* 

Type of Curve 







* From Wyatt et al., op. cit., 1938, p. 30. Reproduced by permission of the Controller 
of His Britannic Majesty's Stationery Office. 

t Calling no boredom 0-1, slight boredom 1-2, moderate boredom 2-3,5, and severe bore- 
dom 3.5-5.0, the curve assessments agree with the boredom assessments in 4 cases out of 6, 
IS out of 20, 22 out of 26, and 10 out of 16 respectively. 

The appearance of a rising, or U-shaped, output curve can there- 
fore be taken as a clue to the existence of boredom of the worker. 
Other factors may, however, produce similar patterns in the absence 
of extreme boredom. Link observed steadily rising output through- 
out each work-spell of shell inspectors, but he concluded that it was 
due to the desire of the worker to reach a certain goal by the end of 
the day, to anticipated bonuses for reaching certain levels of output, 
and, possibly, to the effects of practice and rhythm.® 

In an unpublished study, Cain kept complete records of output 
and other variables for each of sixteen workers over a period of a 
week.^° These workers were all engaged in sewing operations, and 
there was considerable variation in their reports of boredom with 
the work. The individual's work curve varied from day to day, and 
no correlation between curve shape and monotony could be estab- 
lished. (See Figures 22 and 23.) In fact, few of the curves showed 
any of the ''typical" patterns for either fatigue or monotony. The 
output during the last hour of the day might show an increase or a de- 
crease as compared with earlier periods. One factor which was im- 
portant here was the personal goal or quota which the worker had set 
for herself. In this particular plant, a bonus system was in effect, but 
the *'task" was so high that few workers attempted to reach it, and 
therefore the bonus was seldom an incentive. Instead, the workers 
had determined on other bases what they considered to be a normal 
day's work. If they were behind near the end of the day, they 
spurted in order to catch up. If they were ahead, they might even 
cease work altogether, clean the machine, or just potter until quitting 

9 H. C. Link, A practical study in industrial fatigue, Journal of Industrial 
Hygiene, 1919, 233-237. 

^'^ P. A. Cain, Individual differences in susceptibility to monotony (unpublished 
doctoral dissertation), Cornell University, 1942, pp. 58-86. 



[Ch. 9 




1 1 1 I 1 1 1 1 1 1 

t M I I i I t H 1 I I 1 I t I I I 

2.5rOP£/?/^TOR 2/^' 
2.0 - 

-W£SO/^y JUA/£ /6- ''m/?/( /S NOT MONOTOA/OUS^^ 

I I I I I 


I I t I I t I I I I I III I 

WU/?SOt^'/JUA/e/8-"l/VO/?/<i^£RV /Nr£R£SVNG-'^ 



I ' t I I I I I M M I I I 


^spBcmty B£rkV£eN s mo /o /7.m. 


I I I I I I I I I t I M 

I I I I I I I I i I I 

f i I I I I I 

B0R£O JUSr BEFORE ^.30^'' 

M [ t I I I I I I I 

» t I I t 1 I I I 

I I I I I I I 

2.5 r 

/5NTS0 8/7D EXCEPT /^T WE £ND OF WE mY""^ 

1 I 1 I M t I 




n iilU I II 

t I 1 M I I 



9-00 ;o-oo 


m Z-OO 3:0Cf '^■0,0 


Figure 22. Production Curves for a Repetitious Task. Individual Curves Compared 
with Workers' Reports on Boredom 

(From Cain, op. cit., p. 68.) 



6. TUESOr^y 

Z.0 - 

1 .5- 

t I 1 1 1 

I I I 1 1 1 1 1 1 1 1 1 1 

I I I I I I ! 


M I M 

700 800 9. 00 /o oo mo ;a m 2 00 soo '^■ oo 



Figure 23. Production Curves for One Week of Repetitious Work — One Worker 
Who Considers the Work Very Boring 

(From Cain, op. cit., p. 67.) 



[Ch. 9 

time. Factors like this, rather than degree of boredom, seemed to 
be the crucial factors shaping the output pattern. 

Other behavioral indicators have also been proposed, among which 
are variability of performance, frequency of talking, and frequency 
of voluntary rest pauses. Under certain conditions these may be cor- 
related with the degree of boredom of an individual worker, just as 
the shape of the output curve may be. Cain's study of sewing- 
machine operators, however, produced negative results for these be- 
havioral indicators as well. 

It appears, therefore, that the behavior of the worker can be re- 
lated to degree of boredom only under certain conditions. These 
conditions under which the behavioral indicators have some validity 
cannot, however, be specified. It is also possible that these same 
indicators are affected by many other factors in addition to boredom. 
They certainly do not furnish us with trustworthy ''objective" indi- 
cators of boredom. The only method of gauging the tendency to 
boredom in a given worker is to question him as carefully as possible. 

Here the problem is essentially the same as that in any study of 
motives or attitudes by means of interviews or questionnaires. A 
series of specific questions about various aspects of the work is likely 
to give a more valid indication of the worker's boredom than one or 
two general questions. Wyatt and his co-workers devised a stand- 
ardized series of questions which were then scored on an arbitrary 
scale of boredom.^^ This score was used as their criterion in vali- 
dating other indices of boredom. 

The organic changes which are related to boredom are even more 
unclear than the behavioral changes. The most extensive study of 
this problem was that of Barmack, who measured oxygen consump- 
tion, blood pressure, and pulse rate during the performance of various 
''mental" and motor tasks. He searched for possible relationships 
between the reports of the subject at a given time, his rate of per- 
formance, and these organic measures. The relationships, as we 
might expect, were extremely variable. Barmack's conclusions are 
based upon a qualitative examination of the data rather than upon 
a statistical analysis of the significance of the interrelationships. He 
believes there is a trend toward reduced oxygen consumption and 
vascular activity during boredom, and an increase in those functions 
during interested participation in a task. 

11 Wyatt, Fraser, and Stock, op. ciU, 1938, 4. 

12 J. E. Barmack, Boredom and other factors in the physiology of mental effort; 
an exploratory study, Archives of Psychology, 1937, No. 218. 


Individual Factors in Susceptibility to Monotony. — Several re- 
searches have been devoted to the search for the determinant? of 
individual variability in the effects of repetitious work. In most of 
these studies the investigator has been interested in testing some 
theory concerning the nature of the monotony-susceptible worker. 
Miinsterberg, for example, attempted to show that those who have 
difficulty in judging the number of similar items in a list of words 
disliked uniformity in their daily activities.^^ This approach was 
based upon a theory that some persons are much more sensitive to 
variation than others, and that this was an important factor in adapt- 
ing to uniform, repetitious work. Others have considered the ability 
to automatize a task (to make the performance a "habit" and to 
be able to think of other things while working), intelligence, 
introversion-extroversion, the tendency to "perseverate" (difficulty 
in shifting from one activity to another involving reversed motions) 
and so on. 

Most of the results of these studies have provided little support 
for the theories under test, and, more important for our purposes, 
have not provided us with predictive devices which would enable us 
to sort out those workers who would be more adaptable for repetitious 
work. One factor in the above list must be considered somewhat 
more in detail, however, because it has been given a good deal of 
stress by personnel administrators. This is the factor of intelligence. 

There are two kinds of evidence that low intelligence is associated 
with ability to adjust to repetitious work. From some books on per- 
sonnel administration, the reader may get the impression that this 
evidence is conclusive and that intelligence is a primary factor in 
deciding the suitability of a worker for the simpler manual occupa- 
tions. In fact, however, this evidence is by no means so clear cut as 
these textbooks would have us believe. We must therefore consider 
the results more fully. 

One line of evidence is based upon turnover data. In clerical 
occupations, for example, Bills showed that the turnover rate was 
much higher for persons of high intelligence in the simpler levels of 
clerical work than it was for those of low intelligence. The converse 
was also true. Turnover, however, does not measure boredom. It 
may well be that persons of high intelligence leave simpler jobs be- 
cause they are able to find other work which pays higher wages. The 
value of intelligence as a predictor of boredom is therefore left in 

13 H. Miinsterberg, Psychology and Industrial Efficiency, New York, Houghton 
Mifflin Co., 1913, pp. 195-198. 



[Ch. 9 

More direct evidence was obtained by Wyatt, Fraser, and Stock, 
who related intelligence scores directly to degree of boredom with the 
work as reported in a questionnaire.^* Intelligence scores of the 
''most bored" workers and the "least bored" individuals in 
each of several groups are shown in Table 18. The small 

TABLE 18* 

Average Scores in a Test of Intelligence Obtained by Those 
Individuals Who Received the Highest and Lowest 
Assessments for Boredom 

Most Bored 

Least Bored 


No. in 


No. in 

Intelligence Between Groups 

Factory Group 

















Not significant 






Not significant 












Not significant 



















Not significant 







Not significant 






Not significant 

* Ibid. Reproduced by permission of the Controller of His Britannic Majesty's Stationery 

size of each group makes it impossible to interpret the results, 
since the lack of significance may be due only to the small number 
of individuals in the sample. The authors of the study conclude: 
"Although the groups were small, and the differences seldom statis- 
tically significant, yet in each of the ten groups the average score in 
intelligence of those who were most bored by the industrial process 
was higher than the corresponding score obtained by the least bored 
members of the group. For this reason it is permissible to conclude 
that the amount of boredom experienced by operatives employed on 
repetitioa work increases with their degree of intelligence." Ap- 
parently there was so much difference among the different factories 
and subdivisions of the subjects that the authors did not believe a 
combination of all of the resuhs would be any more significant. 

14 Wyatt, et al., op. cit., 1938, 19. 
■^Ubid., 19-20. 



It should be pointed out that prediction of adjustment would re- 
quire more than significant differences. It would be necessary to 
show that there is little overlapping between the groups. For ex- 
ample, there might be a significant difference between the average 
height of college students today as compared with twenty years ago. 
Yet we should not expect to get very accurate predictions of the 
height of an individual merely through knowing that he is a member 
of the present generation of students. 

In addition to the fact that Wyatt, Fraser, and Stock have not 
demonstrated the predictive value of intelligence by these results, 
there are other results which fail to agree even with the mild associa- 
tion which they found. Cain, in another part of the study of sewing- 
machine operators and other employees of a clothing plant, gave an 
intelligence test to twenty-one workers who had been classified ac- 
cording to their monotony susceptibility.^^ This is not enough, of 
course, to establish significant relationships, but trends should show 
themselves, since the individuals represented the extremes of suscepti- 
bility to boredom, no members of the middle group being included. 
Here the difference, although slight and not significant, was in a 
direction opposite to that found by Wyatt, Fraser, and Stock. 

Cain's investigation did provide, however, some clues which may 
lead to more adequate prediction of susceptibility to boredom. She 
administered a questionnaire containing items dealing with personal 
habits, attitudes toward work, home environment, and the like, to 
about seventy workers engaged in repetitious semiskilled work. The 
criterion of susceptibility to boredom was included in the question- 
naire and consisted of questions similar to those used by Wyatt, 
Fraser, and Stock. These criterion questions were mixed in with the 
other items concerning the other personal characteristics of the 
worker. The answers to each item were analyzed separately for their 
relationship to the criterion. Several groups of questions were not 
only related but also showed internal consistency among themselves. 

From this analysis Cain concluded that the monotony-susceptible 
worker tends to be less contented with her life in general than the 
worker who is non-susceptible. The monotony-susceptible also tended 
to be more poorly adjusted in her relationships to family and the 
home. The non-susceptible liked better the more routine kinds of 
housework, and preferred the quieter forms of recreation. She also 
tended to prefer regular habits of daily activity outside of work. In 
other words, there is evidence for generality of the trait of monotony- 

is P. A. Cain, op. cit. 



[Ch. 9 

susceptibility, something which has been frequently assumed but 
seldom demonstrated. 

A system of weighted-scoring of the questionnaire replies, based 
upon the statistical analysis just mentioned, was worked out. It was 
clearly demonstrated that the criterion groups — those used in working 
out the scoring system — could be sharply separated in terms of their 
total scores, with a critical ratio of over 10.^^ This is not yet, how- 
ever, conclusive proof of the predictive value of the questionnaire. 
It is still necessary that the results be checked with a new group 
whose answers were not used in establishing the scoring procedure, — 
a group which answers the questions before employment. It must 
then be shown that their scores correlate with their later adjustment 
to repetitious work. 

The problem of the generality of the trait of susceptibility, touched 
upon above in relation to Cain's study, needs further investigation 
if we are even to develop adequate predictions of adjustment to 
repetitious work. To speak of a trait as general means that a similar 
kind of reaction is observed in the same individual in a variety of 
situations or conditions. Thus, if monotony-susceptibility were 
proved to be general, it would mean that a person who dislikes one 
type of repetitious work would also dislike other jobs of a repetitious 
character, even though the jobs were otherwise quite different. As 
we have mentioned, Cain's results indicated that monotony-suscepti- 
bility on the job is related to dislike for regularity of habits outside 
of working hours. This would obviously indicate generality of the 
trait, but there is other evidence which must force us to proceed with 
some caution in our conclusions. 

Wyatt, Fraser, and Stock arranged to have the same group of 
workers employed in five different tasks in the packing department 
of a candy plant. Each worker spent one month in each of the 
five jobs, and her attitude of interest or boredom was recorded from 
time to time. The results of these ratings are shown in Table 19. 
The numbers in the body of the table represent an assessment of the 
''degree of boredom" determined from questioning the workers, 
with representing "complete absence of boredom" and 5 represent- 
ing ''seldom free from boredom." It will be seen that each worker 
was more bored by some tasks than by others. Wyatt, Fraser, and 
Stock conclude : "The limited results in Table 19 provide some evi- 
dence that boredom may be specific, but a more comprehensive 

1^ A critical ratio of 3 or more is considered highly significant. 
18 Op ciL, 1938, pp. 12-16. 



investigation is necessary before a more definite answer can be 

TABLE 19* 

Showing, THE Incidence of Boredom Symptoms in the Case of the 
Same Group of Workers Employed for Monthly 
Periods in Different Processes 

























. ... 1.7 

Packing Staying 

























Making Bundling 
Crackers Chocolates 























* Ibid., p. IS. Reproduced by permission of the Controller of His Britannic Majesty's 
Stationery Office. 

The difficulty with such an investigation is that it would be neces- 
sary to leave each worker on each job for a considerable period of 
time, and without the worker's knowing that a transfer is to be made 
later. It may be that the one-month period in the above investigation 
is insufficient to determine the specificity of boredom. Cain found 
that in the sewing occupations which she investigated the workers 
who were new at the jobs were relatively free from boredom. After 
finishing the learning period, however, the workers tend to show 
high incidence of boredom. Later, some workers readjust and be- 
come more interested in the work again. Susceptibility to boredom 
should not be measured in an investigation like Wyatt's until this 
period of shifting adjustment is over. The method of selecting 
workers who are not susceptible to monotony will be most satis- 
factory if it will predict which workers will make a long-time 
adjustment to the kind of work. 

From such evidence as is now available, it would appear that there 
are both general and specific factors involved in susceptibility. 
Workers will differ from one another in their ability to adjust to 

19 Op cit, 15-16. 



[Ch. 9 

any kind of repetitive work, regardless of its specific character. At 
the same time, different jobs, all within the class of repetitious work, 
will have varying appeals for the same individual. For the most 
part, predictive methods will be aimed at predicting the general re- 
action to repetitious work. It does not appear likely that it would 
be profitable to attempt to predict differentially for the specific kinds 
of work. That is, it may be feasible to develop a method of pre- 
dicting that a given individual would probably make a satisfactory 
adjustment to the class of repetitious jobs. It would not be feasible, 
ordinarily, to decide whether he would make a better adjustment in 
running a punch press than he would in simple inspection, unless 
there are aptitude differences which are also easily measurable. 
Tendencies to boredom which are related to the specific character of 
the job will probably always be detected by trial and error, except 
possibly in cases where it is especially important that turnover on a 
particular job be reduced to a minimum. 

Wyatt, Fraser, and Stock make the point that a very slight dif- 
ference in the type of work may make a distinct change in the atti- 
tude of the worker. If these slight variations from job to job turn 
out upon more thorough investigation to be much more important 
than the individual's susceptibility to repetition in general, the possi- 
bility of adequate selective devices becomes remote. That is why a 
thorough investigation of this problem is so important. 

Chapter 10 


Even though pay may have been overrated as an incentive to high 
productivity in the past, it remains, nevertheless, one of the key 
problems of industrial motivation. Relative pay is more important 
than the absolute amount of money in the pay envelope, not only be- 
cause of the economic factor of the value of the dollar, but also 
because the pay needs to be adjusted so that the worker feels that he 
is treated fairly in comparison v^ith other workers. 

The worker's pay depends directly or indirectly upon his rate of 
production as compared with some standard of reference. If he is 
paid by the piece, the price paid per piece depends upon what the 
worker's time is considered to be worth, and the standard, or normal, 
speed of work expected from a worker on this job. If he is paid a 
bonus for production in excess of a certain amount, his pay again 
depends upon a standard of reference — how much he is expected to 
accomplish with normal effort. Even if the worker is paid by the 
hour, increases in hourly pay and promotions depend upon whether 
or not his performance exceeds the normal rate for that job. 

The accurate determination of the standard rate or standard time 
for a job is, therefore, an important phase of establishing effective 
industrial incentives. The standard time is also of importance to 
management for other reasons. In planning production, determining 
machinery and labor requirements, or estimating costs, the normal 
time for each unit of work in each job is obviously essential in- 

We need say little at present about the other important phase of 
pay determination and rate setting — that is, of the problem of de- 
termining value of the worker's time when he is working at the ex- 
pected or normal rate for the job. It is evident that this question 
involves many non-psychological aspects, and that the question is 
largely a matter for economic study and political argument. There 
are, to be sure, certain phases of the problem of determining the 
value of the worker's time which are psychological; these will be 
taken up later under the heading of techniques of job evaluation 




[Ch. 10 

in Chapter 11. Even that discussion will not, however, lead us into 
consideration of such questions as whether management and labor 
are receiving their fair share of profits. While the opinions of the 
worker upon such questions will have important effects upon motiva- 
tion, the answers to such questions are matters of economics, political 
philosophy, or ethics, rather than of psychology. 

Even if these broad questions of the distribution of profits were 
answered to the satisfaction of all, we should still be faced with the 
problem of determining what the normal or expected speed of work 
should be for a given task if pay is to be an effective incentive for 
performance. If the expected speed is too high, it fosters resent- 
ment and the worker may not produce as much as he would with a 
more accurate standard. If the expected rate of work is too lenient, 
the worker either fails to do his fair share of work or he receives 
pay which is abnormally high compared to other workers who are 
doing similar work and expending the same effort. If the speeds ex- 
pected by management are equitably determined, there is greater 
chance that the worker will adopt these standards as his own goals. 
If they are not just, the workers develop their own standards which 
may or may not be the most efficient. 

Methods of Determining Time Standards 

Although some of the methods by which standard time is deter- 
mined are spoken of as ''scientific," we shall see that all the methods 
now in use are based upon rule o' thumb and judgment, and that 
many of the techniques do not appear to have any well-defined aim. 
Nevertheless it should be remembered that these methods are all that 
are now available. Their shortcomings merely emphasize the need 
for more adequate research into the problem — research which is 
likely to be slow and very difficult in its development. Perhaps such 
research should be sponsored jointly by labor organizations and 
management in order to insure acceptance of the results as neutral and 

Before evaluating the specific methods, we must consider what it 
is that we want to accomplish. As we have already mentioned in an 
earlier example in this book, the ideal is that workers on different 
jobs, but with the same skill, training, and risk, will receive eqml pay 
for equal effort. General training requirements, the conditions of 
work in the occupation as a whole, risk, and the like, are taken into 
account in deciding the value of the worker's time (job evaluation), 
as was pointed out above. In setting job standards, it is necessary 

Ch. lo] 



to hold skill and effort constant in order to find out how much is 
produced in each kind of work at this constant level. Standard time, 
in other words, refers to the time required by a worker of some 
standard skill and using a standard quantity of effort. 

It may be well to analyze the implications of the preceding para- 
graph in somewhat greater detail, even though at the moment it may 
appear to involve the unnecessary expansion of obvious statements. 
Even though obvious, some of these points have been overlooked or 
confused in the various techniques of time study now in use. First 
of all, we may accept as commonly agreed that a man should be paid 
in accordance with his skill. If he is able to work faster than the 
next man with no greater effort, we consider that he is- entitled to 
the extra gain he can receive in piecework or bonus payments. In 
addition, two men of equivalent skill should, we believe, receive pay in 
accordance with their relative amounts of effort exerted in the task. 

To say that a man is paid in accordance with both skill and effort 
may appear to be no more than the statement that he is paid for how 
much he produces. It is another way of saying it, however, and a 
separation of the factors of skill and effort is necessary if we are to 
adjust the relative pay allotted to different jobs and work under dif- 
ferent conditions. For example, the statement that pay should be 
proportional to output does not tell us anything about the relative 
piece rate which should be applied to two different jobs such as drill- 
ing castings and operating a milling machine. The problem of the 
relative worth of one operation on the drill as compared with one 
operation on the milling machine cannot be solved without some 
analysis of the factors of skill and effort as separate contributors to 

Although skill is a term which is fundamental in much of time 
study, it is seldom clearly defined. We can define thelerm quite 
simply in relation to other concepts which have been used throughout 
this book. An individual's skill simply refers to his capacity for 
performance of a particular kind measured under some standard 
condition. It represents the effect of aptitude plus training and ex- 
perience. There are, however, certain restrictions which must be 
included in the field of time study. If a man is taught an entirely 
new method of performing his work so that he is really not perform- 
ing the same job any longer, his capacity for output may be, of 
course, increased. It is not to be expected that the man should be 
paid for this increase in capacity for performance. If he is the in- 
ventor of the new method he may be reimbursed for his invention, 
but this is a separate problem. 



[Ch. 10 

Some time-study engineers have taken the point of view that any 
change in skill entails a change in method of work, and is therefore 
not to be taken into account in determining pay rates. The point of 
view would be, apparently, that the changes in method could be 
taught to other workers, and therefore would not constitute a factor 
for which a man should be reimbursed in his regular earnings. Even 
if skill were entirely a matter of method of w'ork, which remains to 
be proved, many of the changes would be so subtle that they could 
not be laid down as a part of the instructions for the job. Conse- 
quently such variations in skill must be taken into account. 

We have already discussed the meaning of the term effort in an 
earlier section of this book. (See pages 95-98.) The term appears 
to be used in much the same sense in time study, although its 
meaning is more frequently implied than explicitly stated. 

We may now formulate the problem of time standards a little 
more explicitly. The standard for the job is to be the time required 
for a man of a certain standard level of skill working with a certain 
standard level of effort. Variations in productivity, whether due to 
changes in effort, to changes in capacit^^ (skill), or to changes in both 
factors together are referred to this standard rate of production in 
order to determine how much additional pay should be given for the 
extra output. 

The ideal solution of the problem would be based upon an experi- 
mental analysis of the separate contributions of these two variables. 
The range of skill might be determined for a large group of unse- 
lected individuals who work under conditions of fixed effort-level. 
Some point in this distribution could be chosen as a reference point. 
Then individuals of a fixed level of skill or capacity would be ob- 
served while working at variable levels of effort. Unfortunately, 
such experimental control and sampling is impossible under practical 
industrial conditions, due to the vast numbers of jobs which must be 
standardized and to many other complications which arise. A num- 
ber of short-cut substitutes for the above ideal procedure have there- 
fore been developed by practical time-study engineers. W^e must 
now consider how adequate these substitutes are. 

1. Standards Set by Tradition. — In many occupations which have 
been carried on with little change in method for many years, there 
is common acceptance of a level of performance which is considered 
a normal day's work. The brick layer is expected to lay x bricks per 
day in straight work, the painter can cover about x square feet of 


clapboard wall, and so on. While a well-motivated workman ought 
to be able to determine a reasonable speed of work over a number 
of years at the same kind of work, the worker in the normal working 
situation is not motivated to find the most reasonable rate of work 
(the one which he can maintain without excessive fatigue). Instead, 
he is motivated to find the lowest rates which will be accepted by 
the employer. 

Even if we were willing to accept rates of performance sanctioned 
by tradition as satisfactory bases for evaluating a man's performance 
and setting his pay, this procedure would be applicable only to a 
relatively small number of jobs, such as the standard skilled occupa- 
tions. Large numbers of factory jobs change so rapidly that there 
is no time for a body of skilled workers to arrive at a traditional rate 
of performance. Consequently it is understandable that industry has 
sought something more objective, and something which would per- 
mit a statement of the standards for a job almost as soon as that 
job begins. The other methods' which we shall consider represent 
examples of the solutions which have been widely accepted as the 
best available at the present time. 

2. Time Study with Rating. — In order to understand this widely 
used method, it will be necessary to describe briefly the steps by 
which the time-study engineer arrives at standard time for the job. 
He defines standard time as the time required by a person of average 
skill, working with average effort with an hourly wage, and under 
average conditions, to do the job in question. He also takes into 
account allowances for decrease in rate of work due to fatigue as the 
day goes on, normal interruptions in work such as waiting for more 
material, rest periods, and the like. 

It should be noted at the outset that the term "average" is not 
clearly defined, but that it probably does not mean a statistical av- 
erage. It is, rather, an arbitrary standard or norm by which the 
time-study engineer works, and certainly it does not coincide with the 
average level of performance of workers on the job. Barnes, for 
example, gives a table illustrating individual differences by showing 
the distribution of rates of performance in a group of workers on 
the same job.^ In this group, only four of the one hundred twenty- 
one workers take longer than the standard time. Of course they were 
presumably paid on the basis of a piece rate or bonus, while ''standard 
time" refers to performance without this additional financial incen- 

1 Barnes, op. cit., p. 271. 



[Ch. 10 

tive. It is difficult, therefore, to determine whether the standard is 
intended to refer to a true average under the latter conditions or not. 
At any rate, some caution must be applied to the use of the term 
average in this connection.^ 

The standard time can be determined with the rating or ''leveling" 
method from observations upon a single worker, although the re- 
sults may be checked by observations upon other workers. The 
time-study man first observes the job and lists the component parts 
or elements which make it up, describing them as ''place part in jig," 
"tighten nut," ''push hand lever," and so on. After making certain 
that he has listed all possible parts of the operation, he times a 
worker with a stop watch, with the knowledge and consent of the 
worker concerned. The time for each part of the operation is re- 
corded separately.^ 

If a sufficient number of records is made, it is possible to derive 
the average time for each part of the job and for the job as a whole — 
for this worker, under these specific conditions, with a given amount 
of efifort. It is now necessary to make allowances for the special 
worker and conditions in order to level the times to the standard 
worker and standard effort and conditions. 

The procedure of "leveling" developed by the Westinghouse 
Manufacturing Company is based upon ratings.* The time-study 
man rates the worker's skill, effort, and consistency, and the condi- 

2 Shumard states that production, should average one-third greater under a 
suitable piece-rate system than under a straight time payment. (F. W. Shumard, 
Primer of Time Study, New York, McGraw-Hill Book Co., Inc., 1940.) If this 
is true, average time under piece rate would be only three-fourths of standard 
time, even if workers of "average" skill are employed. If the workers are selected 
for skill, there would be a still greater reduction in average time. The problem is 
complicated by the fact that piece-rate systems will vary in their effect, depending 
upon the type of work and other factors. (Cf. S. Wyatt, Incentives in repetitive 
work, Industrial Health Research Board (Great Britain), 1934, Report No. 69.) 
As the standard time is defined, it is therefore difficult to see how useful it would 
be for work scheduling or other situations where it is desired to predict truly 
average production rates. 

A more extended analysis is given in a recent book by R. Presgrave {The 
Dynamics of Time Study, (2nd. Ed.), Toronto, University of Toronto Press, 1945) 
where it is recognized that the standard is to be adjusted in a manner to make the 
pay incentive maximally effective. Presgrave chooses to adjust the standard so 
that average performance of a group of workers would be approximately 25 per 
cent to 30 per cent above standard when paid on a piece-rate plan. Only about 
5 per cent of the group would fall below "standard." 

3 The reason for timing each part of the job separately, rather than taking the 
over-all time, involves other aspects of time study which we do not need to discuss 
at the present time. 

It is understood here that the proper method of doing the job is already estab- 
lished. See Chapter 7 for a discussion of procedures by which methods are devel- 
oped — motion study or methods engineering. 

4 S. M. Lowry, H. B. Maynard, and G. J. Stegemerton, Time and Motion Study. 
(3rd Ed.), 1940, McGraw-Hill Book Company, Inc., New York, p. 233. 

Ch. lo] 



tions under which the study was taken. These ratings are translated 
into numerical ''leveling factors" which are used to correct the ob- 
served times. Table 20 shows the descriptive ratings which are 
employed in this system, and the corresponding numerical factors.^ 

TABLE 20 * 
Leveling Factors 


+ 0.15 


+ 0.13 


+ 0.11 


+ 0.08 


+ 0.06 


+ 0.03 



















+ 0.06 



+ 0.04 



+ 0.02 













+ 0.13 


+ 0.12 


+ 0.10 


+ 0.08 


+ 0.05 


+ 0.02 



















+ 0.04 



+ 0.03 



+ 0.01 












* Reproduced by permission from S. M. Lowry, H. B. Maynard, and G. J. Stegemerton, 
op. cit., p. 233. Copyrighted, 1940, by the McGraw-Hill Book Company. 

Let us illustrate the application of this rating procedure by a crude 
example. For simplicity, we shall assume that the factors are to be 
applied to the time for the job as a whole, rather than to the individual 
parts of the job. 

^ Other rating methods appear to be simpler than the Westinghouse method, but 
they are actually based upon the same general ideas and are simpler only because they 
are less explicit in stating precisely what is involved. See F. W. Shumard, 
Primer of Time Study, New York, McGraw-Hill Book Co., Inc., 1940, and Phil 
Carroll, j"r., Time Study for Cost Control, New York, McGraw-Hill Book Co., 
Inc., 1938, for examples of other rating procedures. 



[Ch. 10 


To Illustrate the Application of Leveling Factors 

Average time for worker under study 

10.00 minutes 

Ratings of worker 

Skill: Fair (El) 

Effort: Poor (Fl) 

Conditions: Average (D) 
Consistency: Poor (F) 

Leveling factors : 




Total of leveling factors 

Time for average worker (21 per cent deducted from 

10.00 minutes) 

Fatigue and personal allowances 

7.90 minutes 

8.69 minutes 

Standard time (10 per cent added to 7.90 minutes) 

The reader will probably raise at once the question of how ac- 
curate these ratings may be. Other questions will probably occur to 
him — such questions as why consistency should be considered, since 
it is probably already taken into account in rating skill and also 
affects the actual times recorded in the study. Also, how was the 
numerical factor determined for each rated level of skill, effort, or 
conditions ? 

We have chosen to illustrate the details of only one out of several 
methods in use. The proponents of other methods criticize the above 
procedure in terms of certain of its specific aspects, but all methods 
have many features in common. Although the term "leveling" usually 
refers to the method of LowTy, Maynard, and Stegemerton. sketched 
above, we shall find it convenient to refer to all similar procedures 
as "leveling methods," even though the authors of these other 
methods would object that their procedures are distinct from the 
method just described. The reason we feel justified in this grouping 
is because most of the considerations which we shall discuss in suc- 
ceeding sections apply with slight variation to all procedures involving 
rating in finding the standard time. 

As an example of another procedure which is believed by its 
author to be more accurate than the Westinghouse method, we may 
mention the effort-rating method of Presgrave.® Presgrave restricts 
the correction which he applies to time-study data to a rating of 
effort alone. His position seems to be that skill cannot be indicated 
upon a numerical scale insofar as it involves variations in method of 

6 R. Presgrave, op. cit. 

Ch. lo] 



work. Variations in speed which are dependent upon practice or 
aptitude will appear as inseparable from the influence of motivation 
or effort. Consequently Presgrave uses an effort rating which he 
makes synonymous with the rating of ''speed" or ''tempo," and which 
would presumably include any purely quantitative effects of skill. If 
skill entails variations in method, Presgrave believes that it is outside 
the province of time study to settle the issues involved. 

The result of Presgrave's analysis is a confusion rather than a 
clarification of terminology. It may be that in practice his method 
is as reliable or even more reliable than others. That is a question 
which can be settled only experimentally, and in succeeding sections 
we propose to consider questions of the accuracy of time study in 
greater detail. Presgrave's claims for the method are based, how- 
ever, upon its logical consistency, and he presents little concrete data 
upon its empirical accuracy. His method is essentially equivalent to 
the leveling method, since it involves a rating factor for the composite 
which he calls effort, a rating factor which is used to correct or 
"level" the observed time-study data. 

Reliability of Leveling. The answer to the detailed questions 
about the method seems to be that the technique was evolved through 
trial and error, and the various factors adjusted by experience. As 
for the general question of accuracy, direct evidence is scanty and 
unsatisfactory. The problem must be broken down into two main 
parts, just as any problem of measurement, estimation, or prediction 
must be. The first phase of the question is whether the method is 
reliable — will different time-study men arrive at the same standard 
time for the same job ? 

Without giving specific data, Barnes summarized the opinion of 
many time-study engineers on reliability as follows : 

Time standards set by any one of a group of competent and well- 
trained analysts, or by the same one on different days, should fall within 
5 to 7 per cent of the average of the group. That is, no one should differ 
more than 10 to 15 per cent from any other. Some motion and time- 
study analysts claim that they can consistently do this within 2 or 3 per 
cent, but this is doubtful.'^ 

Barnes does not say whether these values which he considers as 
estimates of the accuracy of time study would be obtained if different 
workers were studied in the different time studies being compared, 
or if they apply only to different studies of the same individual. He 

7 Ralph M. Barnes, op. cit., p. 277. 



[Ch. 10 

also does not state whether the analysts were trained together and by 
the same instructors, or whether they learned their techniques more 
independently. With these questions still before us, and with the 
lack of concrete data which now exists, we cannot now draw definite 
conclusions concerning the reliability of the leveling method. 

The question of reliability of the leveling method will need to be 
separated into two parts before a clear answer can be found. There 
is, first of all, the problem of consistency of the method as applied to 
the same worker by different time-study men, or the same time-study 
man at different times. This consistency depends, primarily, upon 
the reliability of the ratings for skill, effort, and the like, although 
variations in the man's performance from day to day might compli- 
cate the problem when the studies are made at different times. The 
generally low reliability of merit ratings is well known, and Uhr- 
brock ^ gives an example of the low reliability of ratings of quality 
of work. These facts should raise serious doubts of the possible re- 
liability of ratings of effort and skill. The time-study ratings are, 
however, different in character from these other ratings, so that 
consistency might turn out to be more satisfactory in the case of 
time study. Direct study is necessary, and the low reliability of other 
ratings emphasizes the importance of such research. 

The second phase of reliability involves the consistency of time 
standards when arrived at through studies of different workers, either 
by the same or different rate-setters. Consistency in this respect 
depends not only upon the consistency of the ratings as discussed 
above, but also upon the correctness of the numerical leveling factors 
assigned to the various men under study. In order to achieve con- 
sistency of time standards under these conditions, the raters must be 
able to place the men in their correct relative positions upon the scale 
of each factor, and the numerical factors permanently assigned to 
each rated level must be correct. In other words, it is not only 
necessary that the same worker should always be placed at the same 
point on the scale of, say, skill, but that two workers who are actually 
20 per cent apart in their skill should always be placed that distance 
apart in terms of the leveling factor assigned. It should be evident, 
therefore, that consistency of standards based upon the same worker 
in repeated studies would not guarantee consistency when different 
workers are involved.^ Uhrbrock has also considered this phase of 

8 R. S. Uhrbrock, A psychologist looks at wage incentive methods, American 
Management Association, Institute of Management Series, 1935, No. 15, 17-20, 

9 The use of the term reliability in this connection may be questioned by those 
familiar with its usage in the field of psychological testing. Reliability involves 


the problem, pointing out the wide variation in estimates of the range 
of individual differences in ability which would make it difficult to 
establish any fixed constants for use as leveling factors. 

These two phases of reliability do not appear to be clearly dis- 
tinguished by many of those who have considered the problem. The 
Society for the Advancement of Management has a committee study- 
ing the whole problem of reliability.^^ Their present plans are con- 
cerned primarily with reliability in judging the same worker, 
although they do mention a possibility that they will also study re- 
liability of times derived from different workers. Other statements 
of the problem, like that of Barnes quoted above (page 217) and of 
Louden fail to point out the distinction. 

The kind of research necessary to evaluate the reliability of time 
standards from this second point of view is self-evident, once the 
question is raised. Evidently we need an extensive collection of data 
upon the Hme standards for identical jobs as established by different 
raters in different plants. With sufficient data of this kind on a 
variety of different jobs, the range of error of the method could be 
clearly established. 

These are questions which cannot be answered in the light of data 
which have been presented thus far. Barnes' statement quoted above 
is apparently an estimate or opinion without concrete statistical evi- 
dence to support it. In fact these questions are seldom even con- 
sidered in writings upon these methods. 

Cohen and Strauss have given us one of the few direct empirical 
studies of this problem of reliability of the leveling method as it is 
applied to different workers. Twenty-one operators on the same 
job were subjected to a motion-picture time study. Their task con- 
sisted of "the folding of an 18 X 18 inch gauge sheet to a size of 
approximately 4X4 inches. This was accomplished in six foldings, 
although micromotion analysis revealed that all of the operators did 

a problem of consistency of result, but in this case it depends upon the correctness 
or validity of certain of the tools employed by the time-study engineer. Whether 
it should be termed validity or reliability is therefore a matter of choice. As 
will be seen presently, there are also other phases of the validity of the method, 
however, and it seems clearer to reserve the term for this later discussion. 

10 Op. cit., pp. 15-16. 

11 J. M. Juran, Progress report of committee on rating of time studies. Advanced 
Management, 1941, 6, 110-115. 

12 J. K. Louden, Management's search for precision in measuring a fair day's 
work. Advanced Management, 1942, 7, 29. Louden suggests using multiple judges 
in order to improve the reliability, and the rejection of standards when the raters 
fail to agree, but he does not specify that the different studies be taken upon differ- 
ent workers. 

13 L. Cohen and L. Strauss, Time study and the fundamental nature of manual 
skill, Journal of Consulting Psychology, 1946, 10, 146-153. 



[Ch. 10 

not make their folds in precisely the same manner." Each operator 
was rated for skill and effort by three trained time-study engineers. 
Cohen and Strauss analyzed the times taken by each operator, and 
then applied the leveling factors corresponding to the engineers' 

If the theory of the leveling method is correct, the "leveled time" 
should be the same, no matter which operator was used for deter- 
mining it. Actually, the leveled times were not even approximately 
equal, as shown in Table 22. The maximum value is more than 
twice as great as the minimum value. 

TABLE 22 * 

Time Study of Operators' Performance in a Folding Job 



Actual No. of 

(L. F. X A. F. T 



Frames Taken 

Leveled Time 

























* Cohen and Strauss, op. cit., p. 148. 

In analyzing the film records, Cohen and Strauss found that the 
faster workers had developed a pattern of movements different from 
that of the poor operators. Thus they were not only faster in per- 
forming their movements, but they also had fewer motions to make. 
This is probably the main reason for the breakdown of the leveling 
procedure. A fixed rating procedure cannot take account of these 
subtle changes, since the opportunities for such improvements will 
vary from job to job. As we have already noted, Presgrave and 
others take the position that rating is not supposed to allow for dif- 
ferences in method of work. The logical outcome of such an attitude 
would be that time study cannot be used reliably until a micromotion 
study has been made of the jobs, and until the method of work has 
been specified in the most minute detail. This is only seldom done in 
practice, and even if it were done, the method of one worker would 
never be exactly like the method of another worker. 

Cohen and Strauss's data apply to only one job. It would there- 
fore be dangerous to conclude upon the basis of this one result that 
all time study is unreliable. Their explanation of the results is. 

14 Ibid., 147. 

Ch. lo] 



however, the sort of variation in movement pattern which could be 
expected in most jobs. We may therefore expect further studies to 
agree with their findings. At any rate, there is little documented 
evidence for the consistency of the leveling method, and here is con- 
crete and detailed evidence against it. 

Relative Validity of Time Standards. A method of setting 
time standards might be consistent for each specific job and still 
give incorrect results. That is, even if studies of different workers 
on the same job lead to the same time standards for that job, the 
standards may be too high or too low with respect to the standards 
for other tasks. This question of the proper relative standards for 
different jobs is referred to here as the relative validity of the method. 
Beyond that there is a still broader question of the absolute validity 
of the standards which will be considered as a separate problem. 

The essential problem in relative validity is whether standards of 
judgment remain fixed from one kind of task to another. It is con- 
ceivable, for example, that what is judged as '"normal" effort on one 
job may actually involve much more cost to the individual than what 
the time-study man considers to be "normal" effort in another task. 
There may be illusions which make a man in one kind of work appear 
to be working harder than he is, and which produce the opposite effect 
in other kinds of work. If these illusions affected all time-study 
men in about the same manner, there would be no indication of their 
effect in the consistency of results. 

If such variations in standard from job to job exist, they probably 
constitute the basis for many of the complaints and criticisms of time 
study by the workers. The possibility of such variation does not seem 
to occur to the proponents of various time-study methods, however, 
so that the problem has not been discussed. 

15 Presgrave, who supports a method based upon rating of effort or "speed" of 
work, gives examples to show what is meant by "normal" effort. (R. Presgrave, 
Effort rating. Advanced Management, 1939, 4, 126-133.) These examples include 
walking at three miles per hour, typing fifty words a minute, and the like. By 
such examples it may be possible to establish more reliable criteria in various raters 
by making the "normal" standard more concrete. How it is determined that walk- 
ing three miles per hour and typing fifty words per minute require the same effort 
is not clear. He does point out that the professional record in typing for one hour 
and the five-mile walking record both represent speeds that are 180 per cent faster 
than the normal speeds given above, and he believes that these agreements support 
the validity of his standards. Presgrave's standards are complicated by the fact 
that he makes no separation between skill and effort, unless skill entails a differ- 
ence of method of work as well as differences in speed of motion. 

Even if we accept Presgrave's standards for walking and typing, these are only 
examples to guide the time-study man. He must attempt to apply these standards 
to many other tasks, and in this application the variations we have suggested above 
could readily occur. 



[Ch. 10 

As far as the present writer can see, there is no way in which the 
relative vaHdity of time standards can be checked until it is possible to 
apply direct measures of effort and cost of work to subjects perform- 
ing a variety of tasks. If subjects on various tasks who are working 
at what the time-study analysts consider ''normal" effort are all found 
to be actually exerting the same amount of effort as directly measured, 
then the relative validity of the standards is established. 

While we can come to no conclusions upon this point, and cannot 
find even suggestive evidence to consider, the problem of relative 
validity must be emphasized. It must be realized that establishing 
the reliability of these techniques is still a long way from demonstrat- 
ing the fairness of these standards. The studies by the Society for 
the Advancement of Management, referred to above, are therefore 
only a first step toward evaluating these procedures. 

Absolute Validity of Time Standards. The analysis of the 
problem of time standards has been phrased up to this point in terms 
of reliability and relative accuracy of the method. There is another 
broad question, still more difficult to formulate and to answer, which 
must be mentioned in order to round out the statement of the prob- 
lem. This additional aspect of the accuracy of time standards might 
be termed the absolute validity of the method.^^ 

Suppose, for example, that it has been agreed (miraculously) by 
all concerned what the average earnings of workers in each type of 
work should be. The standard times determined by leveling might 
bear the proper relation from job to job (relative validity), but it is 
still possible that all of the times are too lenient so that average earn- 
nings on all of the jobs are too high, or the workers are not exerting 
a reasonable amount of effort. On the other hand, all of the standard 
times could be too short, so that the effort would have to be unreason- 
ably high in order to bring the average earnings to the agreed level. 

There is no way to investigate this broader phase of validity by 
analysis of time-study procedures alone. Conclusions upon this 
problem depend upon answering such questions as, ''What minimum 
level of skill is to be considered allowable in each kind of work?" or 
"What level of effort can we reasonably expect of the person of 
acceptable skill in each kind of work?" 

Absolute validity depends upon two general phases of time-study 
technique. First, it depends upon the absolute reference points of 

16 See S. Barkin, Labor views the working day, Advanced Management, 1942, 
7, 32-37. 

Ch. lo] 



"normal" effort, skill, conditions, and consistency adopted by the 
time-study engineer. Even if time-study men can agree upon their 
use of these standards of reference, they are still purely arbitrary. 
For example, what the time-study engineer classifies as ''normal" 
effort bears no necessary relation to the optimal effort for the given 
type of work. In Lowry, Maynard, and Stegemerton's descriptions 
of the levels of effort, average seems to be an artificial mid-point 
between the extremes of "poor" effort and "killing" effort. Average 
effort itself is not described, except with reference to other levels. 
The criteria of poor effort include such descriptions as "obviously 
kills time," and "works slowly and appears lazy." Killing effort in- 
volves a worker who "extends himself to [a] pace impossible to main- 
tain steadily." Between these two extremes, "fair," "average," 
"good," and "excellent" effort are spaced at varying intervals. 

Whether the arbitrary mid-point of this scale can be set up as a 
goal, so that workers who work at this pace can feel that they are 
doing a reasonable minimum of work, is impossible to say. What this 
question involves does not seem to be considered by those who design 
methods of time study. 

To use a concrete illustration, we have referred on an earlier page 
to Presgrave's adoption of three miles per hour as a "normal" rate of 
walking.^^ There seems to be no particular reason for adopting this 
rate rather than two or four as normal, and from the point of view 
of fair relative treatment of various jobs it would not matter which 
level is arbitrarily chosen as standard. Suppose, however, that a man 
is paid on a "piece rate" of $.25 per mile, so that the standard earning 
is $.75 per hour. The man might be capable of walking at an average 
pace of six miles per hour, but would hesitate to double the standard 
earnings. His reasons might involve not only the fear of eventual 
reduction in pay rate if he maintains this more rapid pace, but also the 
feeling that he has "done enough" when he has increased the pace 
by one third or one half. 

Golden and Ruttenberg, who are officers in the C.I.O., present the 
point of view that management has failed to foster adequate morale 
in the workers with the result that the workers have frequently been 
successful in obtaining very lenient time standards. Ruttenberg 
cites one example in which he was able to double the production of a 
department in which management believed there were adequate stand- 

Presgrave, op. cit. 

C. S. Golden and H. J. Ruttenberg, Dynamics of Industrial Democracy, New 
York, Harper & Bros., 1942. 



[Ch. 10 

ards. Of course, any organization which is operating efficiently 
would have a reserve capacity which could be tapped under high in- 
centive. This would not necessarily indicate that the operating speed 
should be increased. When it is possible to more than double the 
output, however, one may wonder whether the normal pace fostered 
by time standards is not too lenient. Golden and Ruttenberg are 
speaking, to be sure, of wartime conditions. Whether they would 
suggest that the pace of production could be permanently increased 
without harm is not indicated. 

The ideal standard for any job should come as near as possible to 
the following requirements : 

1. The standard should be low enough so that most workers who 
are adequately selected for the work could reach standard. If 
the standards are so high that few workers are able to reach 
them, the bonus pay system would have little effect, and workers 
would adopt their own standards of work. 

2. The optimal rate of work from the point of view of worker effi- 
ciency should not be very much faster than the standard rate. 
Since the standard itself serves as a goal and point of reference, 
workers will require very high incentive to go very far above 
standard. This requirement evidently implies a study of effi- 
ciency of performance as a function of rate. 

These requirements are phrased in terms of a standard which is 
considered as a minimum standard, such as most time-study proce- 
dures give us. It might be well to couple investigations of the relia- 
bility and validity of these standards with studies of the incentive 
value of standards. More complete knowledge of how standards are 
regarded by the worker, and how they affect his performance, would 
be very valuable. It might be found, for example, that standards 
would be more effective if stated in terms of an expected average 
performance of adequately selected and trained workers, working 
under current wage incentives. This is in contrast with present 
procedures which give us a standard well below this expected average 
level. It has certain potential advantages which should be explored. 
A worker who meets the standard will know that he is actually per- 
forming at the optimal rate for a worker of average skill. If there 
is any tendency to stop when standard is reached, this would be the 
best rate which could be set. Where scheduling and planning of work 
are based upon standard times, these actually expected times should be 
even more valuable than the present minimal standards. 


Fatigue and Personal Allowances in Relation to Va- 
lidity OF Time Standards. It was noted above that the standard 
time as established by time-study methods includes additional allow- 
ances for fatigue," interruptions, delays, rest periods. Usually, 
the time determined by the leveling process is considered the best time 
which can be made by the average worker ; that is, the time required 
when he is fresh and when he is not interrupted. When that time has 
been estimated, an additional percentage is added to this base. This 
percentage is supposed to correct the standard time to a value which 
would make it representative of what the average worker could main- 
tain throughout the working day, day in, day out. 

Even if we were to accept the whole procedure up to this point, 
the absolute and relative validity of time-study methods depend upon 
the accuracy of this "fatigue and personal allowance." In the more 
painstaking time-study programs, a table of allowances is worked 
out from daylong observations of a variety of jobs. The total time 
spent by the worker in resting or being away from his work is aver- 
aged. The amount of decrement in output late in the working day 
is computed from the continuous series of time studies. From this 
data a table of allowances for various classes of work is evolved. 
Figure 24 is an example of such a table. When additional jobs are 
to be timed, the allowances to be made for fatigue and personal 
requirements are read from the table. Other delays must, however, 
be determined by direct observation. 

Time lost from fatigue and personal needs is highly variable from 
person to person. The aim is to determine an average value. Those 
who require less time for rest or who fatigue less rapidly are then 
given additional rewards for their steady accomplishment. As 
Lowry, Maynard, and Stegemerton point out, stamina and physical 
fitness would be one aspect of the general suitability of the worker 
to the work. These should be rewarded, along with skill and effort. 

There can be little criticism of the general aim of fatigue and per- 
sonal allowances, and there is no question that these allowances should 
vary from one kind of work to another. We are given very little 
statistical data, however, upon the methods by which these allowances 
are determined. It would appear that the allowances are in practice 
closely correlated with the amount of muscular effort involved in the 
job. While decrement, as we have seen (Chapter 3), is probably 
greatest when the work is primarily muscular, the cost of work to 
the worker is not necessarily correlated with the decrement. To make 

^ Lowry, Maynard, and Stegemerton, op. cit., p. 154. 



[Ch. 10 


Handle 70-pound containers from, skid waist-high to shoulder-high stack. 



23 -r 




Handle 60-pound containers from skid waist-high to shoulder-high stack. 
Pull loaded 4-wheel truck under normal conditions. (Gross weight, 2500 pounds. 
Wheel diameter, 11 inches.) 

Up-end rosin barrel weighing 500 pounds gross. (Two men.) 

Shovel salt from open-end box truck to kettle 40 inches high. (Shovel weight, 6 pounds ; 
salt weight, 20 pounds.) 

"Walking on level carrying ?.'> pounds on shoulder. 
Push loaded wheelbarrow. (Weight of material, 350 pounds.) 

Push loaded 4-wheel truck. (Gross weight, 2000 pounds; wheel diameter, 11 inches.) 
Handle 65-pound containers from skid waist-high to R.R. car knee high. 

Handle 40-pound containers from skid waist-high to shoulder-high stack. 

Handle 65-pound containers from skid waist-high to knee-high stack. 
Use pick weighing 9 pounds to loosen new salt in R.R. car. 
Paint smooth ceiling from step-ladder using a 4-inch brush. 

Handle 50-pound containers from waist-high slide to skid. 

Pull loaded 4-wheel truck. (Gross weight, 1500 pounds; wheel diameter, 11 inches.) 

Wet-mop rough concrete floor. 

Dry-mop rough concrete floor. 
Saw a yellow pine 2" X 4" across grain. 
Handle 30-pound containers from waist-high slide to skid. 

Pull loaded 4-wheel truck. (Gross weight, 1000 pounds; wheel diameter, 11 inches.) 
Wet-mop wooden floor in good condition. 
Dry-mop wooden floor in good condition. 

Scrape dirt from wooden floor in good condition. (Handle of scraper 60 inches long, 
blade 6i inches wide.) 
Walking on level carrying 25 pounds. 
Sweep rough concrete floor. 

Handle 20-pound containers from waist-high slide to skid. 
Dry and polish window with rag, working from inside. 
Form and stitch fiber containers. 
Sweep a wooden floor in good condition. 

Wash window with wet rag or sponge, working from inside. 
Pull empty 4-wheel truck, (Weight, 400 pound.s; wheel diameter, 11 inches.) 
Operate typewriter. 

Cut strings on bundles of containers. 
Walking on level unobstructed. 

Visual inspection and maintaining register for printed labels. 

Personal allowance for women. 
Personal allowance for men. 

Figure 24. Fatigue and Personal Allowance Table 

(Reprinted by permission from Motion and Time Study by Professor Ralph M. Barnes. 
Published by John Wiley & Sons, Inc., 1940. P. 2 79.) 

a large "fatigue" allowance for heavy muscular work, and little or no 
allowance for light or "mental" work, may result in demanding a 
higher level of effort for the lighter tasks than that required for the 
heavier tasks. In other words, eyestrain, concentration, postural 

Ch. lo] 



strains, and monotony of the lighter type of work might cost the 
worker more in the long run than the heavier physical effort which is 
given a liberal fatigue allowance, even though the lighter work does not 
show a clear-cut decrement as the day progresses. Once more we see 
the need for facing the problem of cost of work in its broader aspects. 

Proposals to Avoid Rating. Some union organizations have 
made a proposal which is apparently designed to overcome some of 
the possible errors of the rating method. They suggest that union 
contracts specify that an average man should be selected for the time 
study, presumably eliminating the need for a leveling factor for skill. 
This proposal is unsatisfactory in several ways, and certainly it does 
not necessarily work in favor of the worker. In the first place, rating 
for skill is still involved in the selection of the worker. By requiring 
that the selection meet the approval of the workers, the rating no 
longer depends upon the time-study man's judgment, and the workers 
apparently feel that this would result in a lower standard. This is 
not necessarily true, however. Suppose, for example, that the workers 
in a certain job have been carefully selected so that the skill level is ex- 
tremely high. All the men should receive above-average pay under 
these conditions. Yet if the skill rating is eliminated, and if an 
"average" man is selected from the group now working on the job, 
the men on that job would receive average pay for that kind of work, 
regardless of their high standards of skill. Whether or not such a 
situation would often occur, lack of definition of what constitutes an 
average worker would make consistency of time standards impossible. 

Not only does the selection of an average worker for time study 
fail to eliminate the real problem of skill ratings, btit it also fails 
completely to take into account the equally essential rating for effort 
which would apparently still remain in effect with the problems of 
reliability and validity still unsolved. 

Another proposal is to return to the use of "average" in a statis- 
tical sense. The average would not be an arbitrary standard of the 
time-study man, but would be the actually observed mean performance 
in a group of workers on the job. This proposal is essentially a return 
to tradition or experience as a method of establishing time standards, 
with the additional feature that the experience which is used as a basis 
for the fixing of times would be more carefully analyzed and would 
probably be more accurate. 

Such a procedure cannot be regarded as a substitute for a rating 
method, however, since it fails to accomplish some of the things which 
a rating method is designed to do. Taking average times from 



[Ch. 10 

workers now employed, and under present conditions, merely fixes 
present levels of skill and effort as standard. Yet there may be, at 
present, very poor selection of workers or workers who are excep- 
tionally high in skill, or the effort level may be very low as a result 
of restrictive practices, and as a result of the knowledge that rates 
are to be fixed by current performance. Also, it would no longer be 
possible to set rates upon jobs employing only a few workers, nor 
would it be possible to determine rates for new jobs as soon as they 
begin. Since a rating method is aimed directly at overcoming all of 
these difficulties, the proposal to use actual average performance as a 
standard is merely to abandon the aims of the leveling procedure alto- 
gether.^^ This would not be justified unless it had been demonstrated 
that the rating method was of such low reliability that it really was 
not accomplishing what it set out to do. Even then, some method 
for arriving at these goals should be sought. 

The statistical average might be employed in research studies to 
improve the rating procedures, even though they could not be used 
directly in time study. For example, an experimental study of a job 
might be carried out under artificial conditions in which the subjects 
are provided with very intense motivation so that they work nearly 
at their maximum possible rate. In this way, effort would be held 
reasonably constant at a high level. Then the distribution of per- 
formance in various workers would represent varying degrees of 
skill, and it would be possible to obtain an average time for the given 
high effort level. It would also be possible to determine the range 
of variation of skill by studying the dispersion of the distribution. 
This would, of course, be valuable information for verifying the 
leveling factors for skill, although a great variety of jobs would 
have to be studied in this manner, with a large number of workers in 

2 I. Lorge (A statistician looks at time study, motion study, and wage incentive 
methods, American Management Association, Institute of Management Series, No. 
18, 18-26) appears to advocate the use of statistical averages in determining stand- 
ards, but at the same time to retain the leveling procedure. This seems to be a 
contradiction, or at least an ambiguity in his proposal, as the above discussion would 

21 W. Bloch (Die Anwendungen der Mathematik in Zeitstudienwesen, 7th In- 
ternational Management Congress, Washington, 1938, Production Papers, 144-151) 
proposes the use of the average time plus a correction factor based upon the stand- 
ard error of the mean. He would apparently apply this procedure either to a single 
worker or to groups of workers, however, and his discussion does not appear to 
take into account the need for leveling for skill. The correction factor he proposes 
is stated to be a correction for "intensity" of work or effort. The dispersion of 
times increases as the effort decreases. Since the standard error of a mean depends 
upon both the dispersion and the size of the sample, he chooses to take this as the 
corrective factor. Even if this assumption could be fully accepted, there is still no 
reference to variations in skill. 

Ch. lo] 



each task, in order to investigate the variations in range of skill from 
job to job. 

The study of variations in effort could be based upon a prior study 
of skill like that described in the preceding paragraph. Once workers 
had been classified into groups of equivalent skill by the procedure 
described, the workers could then be studied while working at various 
|k submaximal levels. Such study might take a variety of directions. 
B More exact criteria of the various levels of effort might come from 
pi descriptive studies. For example, the variability of times, the fre- 
quency of errors or false movements, or other objective factors might 
show a relation to the effort level. Effort level itself would be estab- 
lished objectively as it has never been done in time-study work, 
because the capacity of the individual worker would already be estab- 
lished with some accuracy. In addition, if long-term physiological 
investigations could accompany such experiments, some hints as to 
the level of effort which is optimal might be produced. If optimal 
effort is once established, the problem of absolute validity of time 
studies can be solved. 

All these suggestions are related, however, to the use of statistical 
averages in fundamental research for improving time study. For the 
reasons outlined before, averages taken in the ordinary conditions of 
work would not solve the problem of time standards, since their use 
would merely avoid the issues by ignoring them. 

It should be said that there is some justification for the use of 
rating methods, in spite of all of the open questions. The developers 
of the various methods have replied to criticisms by stating that rating 
in some form is the only reasonable method now available, and that 
time standards must be determined somehow. There can be little 
quarrel with this attitude. The danger lies with those users of ratings 
who regard them as final solutions, and who forget that they are no 
more than stopgaps. Even though rating must be used for lack of 
anything better, we should keep before us its provisional nature, in 
the hope that industry will provide opportunity and support for re- 
search designed to test and improve the technique. 

3. Synthetic Time Standards — We have given such a lengthy 
consideration to the rating methods of time-study because of their 

twide use, and also because it brings out the fundamental problems of 
all methods of establishing time standards. As a result we can now 
deal with the '*S3^nthetic" method in a small space. 
The synthetic method consists in using data from previous time 
studies to establish the time standard for a new job without using the 

Amount of positioning 


(Medium) 3 F 

1 nrcc iirnjers anu iiiuniD 

(Large) H 
Extended hand 

Poaitionino is normally little more 
than releasing the object on the 
worh place. 

Time Class One Hand 8= 0.006 
No. •! One Hand 12=0.007 
1 Two Hands Sim. 12=0.011 

Time Class 
No. ^ 

Place toaster 

One Hand 6--C.t;:6 
One Hand :itl 
Two Hands Son. L2 = 0.Cil 

Place small onr blank io oihe; hand. 

body in other hand for ajbsc;i,cnt 

Positioning of parts on or into 
dellnlte locations with ample 
tolerances, simple open nests or 
fiitures. or assemblies with one 
point of location. 

Time Class One Hand 8=0.006 
No. "I One Hand 12=0.007 
1 Two Hands Sim, 12=0.011 

Oispose of screwdriver into funnel-type holder^ 
6 inches to right ol work place. ^ 

^"Time Class One Hand 8 = 0.011 
No. O One Hand 24"=0.0:3 
^ Two Hands Sim. 24 = U. 120 

Dispose of toaster assembly to irty. 

Positioning ol parts on or into 
dilticult or complicated locations, 
i Assemblies or fixtures requiring the 
positioning ol parts with respect to 
two definite points, or location in 
two directions. 

Time Class One Hand 8 = 0.011 
No. <<> One Hand 24 = 0.013 
^ Two Hands Sim. 24 = 0.020 

Position In two ^ ■ 
Place screwdriver wrench over standard nut^ 

^ Time Class 



Place power screwd 

One Hand 8 = U.U'9 
One Hand 24 = n.o2l 
Two Hands Sim. 24 = U.0Jl 

[ J \ Position in iwo 
<T)*2^ directions 

river on head ol seFf-tappinfl 

Positioning Is much the same as 
Condition C but in addition may 
involve close tolerances, greater 
care of finishes, three or more 
points or directions ol location, or 
applicition of iorce to assemble. 

Time Class One Hand 8 =^0,019 
No. O One Hand 24 = 0.021 
W Two Hands Sim. 24 = 0.031 


Position In three W 
direction* <L^3 

Place screwdriver on screw in assembly^^ 


This comhiinllos not osed. 

Figure 25. Standard Time Values for "Place." 
(Reprinted by permission from Barnes, op. ext., opposite p. 345.) 


(Small) 2 F 
Two fingers and thumb 

(Very Large) 2 H 
Two hands 





Time Class 
No. 1 

One Hand 8==0.006 
One Hand 12=0.007 
Two Hands Sim. 12=0.011 

Time Class 

Two Hands 8^0.011 
Two lUods 12=0.013 

Place small machine bolt in other hand. 
Rarely used other than for placing part into 
other hand. a 

Time Class 
No. O 

One Hand 8=0.011 
One Hand 24 = 0.013 
Two Hands Sim. 24= 0.020 

Move wallle iron partial assembly to rouoh 

positron under power screwdriver. 


ime Class 


Two Hands 8=0.019 
Two Hands 24=0.021 

Place flat steel washer over stud or pin where 
tolerances are large. V 

Time Class 
No. 3 

One Hand 8 = 0.019 
One Hand 24 = 0.021 
Two Hands Sim. 24=0.031 

Place screw in tapped hole. 

Time Class 


One Hand 8-0.024 
One Hand 24=0.026 
Two Hands Sim. 24^0.039 

Plate nut on terminal in limited space where 
•inoers are cramped by design. 

Place rectangular aluminum casting in flxtoie. 

Time Class Two Hands 8'^= 0.030 
ho. |± Two Hands 24=0.036 

Place toaster fixture on locating pio tox 
driving screw. 

^ Place unit cover plate over unit on walDe* 
A ' Iron grid assembly. 

Corrected for Transport Distances. Time in Minutes. 










[Ch. 10 

stop watch. The procedure is based upon the assumption that any 
task can be divided into a number of component parts or elements, 
and that these elements will be identical with elements to be found in 
other tasks. If all jobs are regarded as made up of combinations of 
a limited number of basic movements, the standard time for the job 
is determined by adding the previously established standards for each 
of the elementary movements which compose it. 

Figure 25, taken from Barnes, illustrates the kind of data which 
are prepared for use in establishing synthetic time standards. Various 
conditions of placing a tool or part are classified, and the time for an 
element of placing which is involved in a task under study is read 
directly from the chart. After the times for all elements of the job 
are added, a fatigue and personal allowance is added to the total to 
give the standard time. 

This method does not require any stop-watch study or rating in 
its application. Nevertheless, the rating is not actually eliminated 
from the complete procedure. The tables of standard times for each 
element were themselves originally established by the leveling method 
or some equivalent procedure. From this point of view, then, the 
only advantage for the synthetic method is that the rating is done 
once for all, and not all time-study men would need to be trained in 
rating. If the original tables of elemental times were worked out by 
a small number of highly trained raters, the inaccuracies due to un- 
reliable ratings could thus be more easily reduced. Rating is still 
present, however, and no one knows how much more reliable the 
procedure might be. 

In addition to the fact that the synthetic time method does not 
avoid the problems of leveling, it has certain additional drawbacks. 
While the errors of rating might be smaller with this system, what 
errors are made will be perpetuated, and will not vary at random 
from job to job as they would in the leveling procedure with stop- 
watch observations of each job standardized. The only hope is that 
errors in the standards for one element of a job will be canceled by 
opposite errors in other elements of the job. 

The fundamental assumption used in synthetic methods is itself 
open to question. It is assumed that the time for a given motion 
should be the same, regardless of what other movements are included 
in the task. It would be just as logical to assume that the time for a 
given movement depends upon both the preceding and succeeding 
elements of the task. Such an assumption would, however, make the 

22 Ralph M. Barnes, op. cit. (opposite p. 345). 

Ch. lo] 



determination of synthetic times an impossibility. Thus the establish- 
ment of this method as an accurate device depends not only upon 
research into the accuracy of the leveling procedure, but also 
upon research into the additional assumptions required by the 

One industrial engineer has carried this assumption even further 
by stating that the time for a simple movement is not only independ- 
ent of other movements in the pattern, but that it is constant from 
individual to individual. As Segur states his principle, ''Within 
practical limits, the time required for all experts to perform a true 
fundamental motion is a constant." In demonstration, Segur 
predicts the time required for an individual to write his signature at 
top speed by analyzing the movements required. 

The implication of Segur's ''principle" would be that all persons 
doing a given task at maximum speed and with the same method 
(i.e., exactly the same pattern of motions) would perform the task 
in the same time. In a letter to Gordy,^* Segur states that the "prac- 
tical limits" would be a maximum variation of 20 per cent. If this 
notion were accepted, the synthetic determination of standard times 
would be based upon times for "fundamental motions" (smaller sub- 
divisions of the total activity than those illustrated in Barnes's figure 
on pages 230-231). The establishment of these times would require 
time studies under conditions of maximal effort or with a leveling 
'factor for effort. No allowance would be made for skill, however, 
since the operator would have sufficient practice to become an "ex- 
pert." The time for the total operation would be established for 
those movements which, in the opinion of the industrial engineer, 
make up the best method of doing the job. In practice, some account 
is taken of the combinations of movements involved in the job, since 
Pollock mentions the need for considering the overlapping of mo- 
tions where the operator prepares for the next movement while fin- 
ishing the current action." In such cases, a combination-time is 
used, but evidently there could not be allowances for a great variety 
of interdependence of movements. 

In support of Segur 's method. Pollock, an employee of Segur, has 
presented several experiments in which the time for an operation was 
predicted according to Segur's method, and then the times required 

2 3 A. B. Segur, Labor costs at the lowest figure, Manufacturing Industries, 1927, 
13, 271. 

24 C. B. Gordy, Variations in the time required for skilled operators to perform 
a simple motor task, American Journal of Psychology, 1943, 56, 181. 

2 5 K. W. Pollock, The use of therblig times for rate setting, Advanced Man- 
agement, 1937, 2, 35-40. 



[Ch. 10 

by a few operators were obtained by direct analysis of films. In this 
rather small number of instances, the total times for the operations 
agreed closely with the predicted totals. The errors in the individual 
elements are much larger, however, and it would appear that the 
small error in the total is due to the random variation in these errors 
from element to element, so that most of the error is canceled out. 

Gordy has put Segur's basic assumption to a direct test." He 
carefully measured the times of over thirty operators performing a 
simple manual task requiring picking up and placing of blocks in a 
container. All the subjects were industrial employees in manual 
work, and all practiced the experimental task until they no longer 
improved before the times were recorded for analysis. They all used 
the same method or pattern of movements, so far as could be de- 
tected by the naked eye, although slight differences appeared when 
the motion pictures were studied. It would appear justifiable, there- 
fore, to consider these subjects as "experts," and a» using substan- 
tially the same movements. The results of Gordy 's experiment are 
shown in Table 23. 

TABLE 23 * 

Mean Cycle-Time, Standard Deviation, and Range 
(In "winks" or 1 /2000th of a minute.) 

Right Hand 

Left Hand 



Standard Deviation 



Range of Individual Times 

669 to 1133 

702 to 1118 

* Ibid, ut supra, 192. 

It is evident that the range of times is such that the slowest 
operator requires about 70 per cent longer than the fastest. The ex- 
treme range could be more accurately estimated from the standard 
deviation multiplied by six. In both the left-hand and right-hand 
data, this would give a range of variation of over 30 per cent above 
and below the mean.^® This is definitely greater than the range 

26 O/J. cit. 

2 7 C. B. Gordy, op. cit., 181-194. 

2 8 Gordy confuses the issue by calling attention to the standard error of the 
mean. This is the measure we would use if we wanted to predict the «u'a;i per- 
formance of another group of operators on the same task. Gordy states this fact 
correctly, but implies that it bears upon the validity of Segur's assumption. Segur's 
statement, however, has to do with individual variation, not with the variation of 
group means. In addition, the standard error of the mean could be reduced by 
merely increasing the number of operators in the experiment, thus attaining as 
small a standard error as is desired. 

Ch. lo] 



claimed by Segur. Even if we were to take twice the standard devia- 
tion as the practical range, thus bringing the result closer to Segur's 
claim, the data represent a very wide variation. We may therefore 
wonder what value a ''practical limit" of 20 per cent might have in 
establishing fair time standards. 

Since the subjects of Gordy's experiments were highly motivated 
and placed on their mettle by the tests, ther^ can be little question that 
differences in skill are reflected in the speed of individual movements 
as well as in differences in method of work. Thus a leveling pro- 
cedure must take account of both skill and effort or of some com- 
posite of the two factors. This point has bearing upon time-study 
method in general, as well as upon the specific procedure by which 
Segur established his tables of standard times for each element of 
motion. Pollock's data is not extensive enough to establish the valid- 
ity of times estimated upon the basis of Segur's method, and Gordy's 
experiments show very conclusively that the method is based upon 
faulty assumptions. 

This conclusion concerning Segur's "principle" does not, of 
course, invalidate all procedures of developing synthetic time stand- 
ards. Even Segur's method might work, in practice, in spite of its 
faulty theoretical basis. In the last analysis, the value of any method 
of establishing synthetic times depends upon the validity of the time- 
tables which are used, and upon the validity of the fatigue or other 
allowances applied to the job. These basic data cannot be obtained 
until the fundamental problems of time study in general are solved. 
When these problems are solved with reasonable accuracy, synthetic 
methods may be a useful short cut, if it can be shown that the addi- 
tion of elements with little regard to combinations and interrelations 
of movements is justifiable. Meanwhile the use of synthetic time 
procedures would appear to be a process of pyramiding methods of 
unknown validity, one upon another. 

Summary. — In order to understand the operation of financial in- 
centives, it is necessary to understand current practices which affect 
the accuracy and fairness of wage rates, particularly the relative pay 
of various workers. We have seen that time study plays an im- 
portant role in establishing the financial incentives for many jobs. 
There is a great deal of mistrust of these methods on the part of the 
workers. While this mistrust is usually based upon false premises, 
and upon faulty understanding of the aims of time study, this mis- 
trust is a potent factor in industrial motivation. Mathewson's 



[Ch. 10 

studies of restriction of output, mentioned in Chapter 8, bear 
this out. 

If time-study methods are to be used at all, the mistrust of the 
workers must be overcome. The most satisfactory way of over- 
coming this mistrust would be to produce concrete proof that the 
methods are reliable and valid, since there could be little criticism of 
the expressed aims of time study. The mistrust is based upon the 
belief that time study fails to accomplish its aim of providing equal 
pay for equal effort, skill, and risk. 

Unfortunately, time-study men have, in the past, shown little 
interest in providing definite statistical evidence of the effectiveness 
of their methods, and the methods have been accepted largely on 
faith. There is some indication that this condition is changing, but 
at present there is still little information available. 

In the present chapter we have analyzed the assumptions under- 
lying the various steps of time-study methods now in use. In doing 
so we have also tried to show what kind of research is necessary for 
evaluating these assumptions and their application in practice. On 
the basis of the large number of unchecked assumptions and pro- 
cedures involved in current practices, it does not appear that a very 
convincing case can be made out for the accuracy of either relative 
or absolute standards of performance now in existence. Until re- 
search has provided a more certain basis for time standards, the ef- 
fectiveness of financial incentives will continue to be variable and 

It will be seen that the concepts of effort and efficiency of perform- 
ance enter into this discussion in two ways. We began our discussion 
of incentives as a consideration of factors which affect the effort level, 
and consequently the efficiency, of performance. We now see that 
the establishment of adequate financial incentives must be based upon 
fundamental knowledge of effort, and upon adequate means of es- 
timating or measuring effort. 

Chapter 11 


We come now to two further procedures in industrial practice 
which affect the motivation of the worker. Job evaluation, as we 
have already seen in our discussion of time standards, is concerned 
with the determination of the basic wage to be paid for a given job. 
Time standards are supposed to tell us how much the ''average" man 
should accomplish on the job. Average production on one job is 
worth more than average production on another, however, because 
one job may require more expensive training and education, requires 
the worker to take greater risks, or to work under more disagreeable 
conditions of heat, cold, or dust. Job evaluation, then, is the process 
of scaling jobs with respect to one another, taking into account all of 
the surrounding conditions and requirements of the job. It is an 
evaluation of the job rather than of men, since individual variations 
in productivity are held constant by basing the evaluation upon the 
standard man or norm for that work. Variations in pay, dependent 
upon individual performance, are then superposed upon the basic 
pay level derived from the job evaluation. 

Merit rating serves a function similar to time standards or piece- 
work prices in its effect upon financial incentives. In those jobs 
where objective production rates are inadequate for determining the 
value of a given worker in comparison to others on the same job, or 
where production rates cannot be easily measured, the value of the 
man is estimated by the supervisor. This estimation is the merit 
rating, also sometimes called the ''efficiency" rating. 

In many jobs, then, establishment of fair relative pay rates for 
men on the same kind of work depends upon the accuracy of merit 
rating, just as fair pay depends upon time-study methods in many 
other tasks. In both cases, the fairness of the monetary incentive 
also depends upon the proper relation between the basic levels of pay 
in different jobs and classes of work. If an assembler is paid less 
than a certain machine operator, when the assembler is certain that 
the jobs are of equal value, the effect is obvious. It is the task of job 
evaluation to make certain that such differences are justified. 




[Ch. II 

As the last example indicates, it is necessary not only to have 
accurate and justifiable methods of attaining these ends, but also that 
they be convincing to the worker. The first step, however, is to de- 
termine the accuracy of the methods. If that is done, the process of 
convincing the worker would consist only in making clear to him the 
basis of the decision. If the methods themselves are unclear and of 
unknown accuracy, it is impossible to convince the worker that the 
results are justified, except by evasions and deceptions. 

Job evaluation and merit rating are grouped together not only be- 
cause of their bearing upon worker motivation in industry, but also 
because the basic techniques commonly employed for both of these 
purposes have many common elements. Both consist of ratings, or 
judgments, and in methods of combining judgments of a variety of 
factors into a single index. The only difference is that one involves 
rating of a job; the other, rating of a specific worker. 

Merit rating is a factor in many other phases of motivation, as 
well as in establishing the pay rate of an individual. Since it is a 
general device for determining the value of the individual to the or- 
ganization, it enters into decisions upon promotion, transfer, and 
layoffs, as well as pay-raises within a single job classification. 

Merit Rating 

Since merit rating is probably more familiar to most readers, the 
general problems of rating will be discussed first in their application 
to this field. Any case in which a foreman or manager decides that 
a given individual is of more value to the company than another in- 
volves a form of merit rating. In recent years this process has been 
systematized and standardized in the hope of securing more reliable 
and valid results. Most of these procedures consist of developing a 
standard rating-form which serves as a record of the foreman's judg- 
ment and is designed to aid him in making accurate judgments. 

Various kinds of rating form have been used, but the one which 
is probably most widely known in industry is the one which we shall 
examine in some detail. Since most of the questions which arise with 
respect to this particular method are also applicable to the other 
methods as well, we shall lose little in generality by limiting our con- 
sideration to this one technique. The kind of rating which we shall 
examine in some detail is the graphic rating method with a system of 
point scores. 

In this method, the designer of the rating- form decides that he 
wishes or needs to have the men rated upon a certain list of character- 

Ch. ii] 




Name . Dept f^'" i^*** 

Employee's Position )ob Class 

Note This rating will represent in a systematic way your appraisal of the employee in terms of his ACTUAL 
PERFORMANCE ON HIS PRESENT JOB In the interests of furthering careful analysis, the follow- 
ing suggestions are offered regarding the use of this form 
I Consider only one factor at a time 

2. Study each factor and the specifications for each grade 

3. Review upon completion to see that the rating of each factor appliei exclusively to the lndl« 

4 Comment fully at bottom of page and on reverse side upon any matter which in your opinion 
needs explanation 






of this job 


of this job 


of this job 

Decs Net 

of this job 



Economy of WjICfolj 
Economy o( Time 'hi* own and 













PfOductiv* Output 














Follows Initfuctionl 

Punctuality and Attendanc* 
Safety Habitt 



Dependable ir< 
most respects 









Attitude Towards the Cpmpany 
Attitude Towards Supervision 
Co-operation with Fellow- 

Inspires othen 
to work with 
and assist 


Quick to 
volunteer to 
work with 
and assist others 


works well 
with and 
•ssists others 


works well 
with or assists 


Does not 
work well with 
or assist 



Figure 26. Example of a Typical Merit Rating Form 

istics — e.g., ''dependability," ''quality of work," "knowledge of the 
job," and such. He then draws up a chart like that shown in Figure 
26, upon which the foreman records the ratings in each trait for a 
given individual. The individual is then given a point score for each 
trait, and these points are added to give an over-all merit score which 
determines his pay or promotability. 



[Ch. II 

Such a numerical score is presumed to be more accurate than 
traditional unsystematic and qualitative verbal descriptions such as 
''he is a very valuable man," ''I would like to see him promoted to 
supervisor," and the like. The use of a merit rating-form is also 
believed to have other advantages besides that of increased accuracy. 
It provides a concrete record of the opinion of the supervisor from 
time to time. It requires him to consider all his men, and to consider 
a variety of qualities. 

The rating can also be used for interviewing workers, explaining 
why they were not given raises, what additional training they should 
have, and so on. It is also suggested that it helps to train the super- 
visors to judge men.^ 

Reliability and Validity of the Graphic Rating Method.^ — The 

value of merit rating stands or falls, however, upon its accuracy. 
''Objectivity" of the record is of no advantage if you are merely re- 
cording unreliable impressions. It does no good to explain to a 
worker that he was not promoted because of certain ratings if you 
have no evidence of the validity of those ratings. 

We are asked to accept the rating-scale technique largely on faith, 
since there is no way of testing its validity directly. The rating scale 
is ordinarily used in situations where there is no other way of obtain- 
ing an estimate of the worker's characteristics except, possibly, for 
records of his production. Even where output records are available, 
it is apparently believed necessary to know more than the productivity 
of the worker, else the rating scale would not be proposed. As a 
result of the impossibility of direct validation, we are told what the 
rating scale is expected to do or what it ought to do, with little or 
no support for these claims. 

In the following discussion we shall consider these supposed ad- 
vantages of the graphic rating scale in some detail. We shall try to 
show that these claims are by no means easily acceptable, in fact, 
that some of them are logically contradictory. Throughout this dis- 
cussion we shall assume that the merit-rating system is satisfactorily 
administered, with proper instruction and training for the raters. 
The difficulties we shall discuss are inherent in the technique itself, 
even when it is administered under the best possible conditions. How 
often these conditions are met is another question, although some 

1 J. Tiffin, Industrial Psychology, New York, Prentice-Hall, Inc., 1942. pp. 
234-238, gives an excellent statement of these potential advantages of rating methods. 

2 This analysis of the accuracy of rating and the proposal for another approach 
appeared in slightly different form in the Personnel Journal, 1945, 24, 6-15, under 
the title, "Merit rating criticized." 

Ch. ii] 



may admit that these rating scales are only too seldom used according 
to the plans of those who designed them. 

Although it is not always explicitly stated, the graphic rating- 
form in such frequent use apparently aims to secure several advan- 
tages over unsystematized procedures which have been prevalent in 
the past and continue to be common. These expected advantages 
might be listed as follows : 

1. Requiring the rater to consider a number of different "traits" or 
aspects of the man rather than a single total evaluation without any 
explicit analysis. 

2. Defining these traits for the rater so that there is less likely to be 
disagreement upon the meaning of terms. 

3. Emphasizing that individual differences involve a number of levels 
along a continuous scale of each quality. Thus it prevents such vague 
terms as "dependable," "skillful," "honest," being applied to a worker, 
and requires instead an estimate of the degree of dependability, skill, or 
honesty possessed by the individual. 

4. By giving examples or definitions, it is hoped that the raters will 
agree not only upon the general meaning of the trait under consideration, 
but that they will also be able to use a common standard for determin- 
ing the various levels of the trait. If this purpose is attained, "average 
skill" would mean about the same thing to various supervisors. (Not 
all graphic rating-forms make the attempt to define levels of a trait, and 
some even fail to define the traits themselves. We are considering here 
the more carefully worked-out rating techniques, and all of the possible 
advantages which might accrue through the use of all known precautions 
in composing the form. There are methods for "correcting" for errors 
due to failure to secure common standards of raters. These will be dis- 
cussed later.) 

5. By deciding upon the relative importance of the various traits 
considered in the rating, and by assigning point values to each level of 
each trait, it is possible to produce a total score which would represent 
the over-all value of a particular employee and would provide ready 
comparison with other employees. 

It is not usually recognized that the first and last of these expec- 
tations are incompatible with each other. If we are successful in 
getting the rater to consider a number of distinct traits and phases of 
the worker, there are several difficulties and a definite fallacy involved 
in applying a total numerical score to the result. The fallacy con- 
sists in the assumption that it is possible to substitute one trait for 
another. The use of a total numerical score means that the worker 



[Ch. II 

who IS weak in skill can make up for that deficiency by getting an 
especially high rating in such a trait as dependability. Perhaps such 
compensation is possible within limits, but in cases where there is 
considerable divergence between the traits of a given individual, his 
total numerical score may not represent at all his relative value to the 
company. In the following example (Table 24) we shall for sim- 
plicity's sake suppose that each of three men is rated on three traits. 
They all receive equivalent total scores, but one would scarcely say 
that they are equivalent for purposes of transfer, promotion, or even 
for establishing relative pay. 


Hypothetical Ratings of Three Workers with Equivalent 
Total Scores 

Point Values * of Ratings for 
Trait Worker A Worker B Worker C 

Skill 10 5 5 

Cooperativeness 5 10 5 

Capacity for future development 5 

Total 15 15 15 

* It is assumed that all three traits are of equal importance (i.e., each is allotted the 
same number of points). 

H. M. Johnson has pointed out this fallacy in connection with 
aptitude test-batteries which take into account several distinct apti- 
tudes.^ As an example, he describes the score which he might receive 
as an operatic tenor, based upon several component abilities. It is 
possible, he says, that his score for aptitude in this profession might 
be fairly good in spite of the fact that he is totally unable to produce 
a high note, and would therefore never have any chance of succeeding 
as a tenor. 

Johnson's discussion was written primarily for psychologists. 
The point has received little consideration in discussions of merit 
rating which are aimed at the practical administrator, in spite of the 
probability that the fallacy is even more important in ratings than 
it is in the field of aptitude testing. In rating, we are not even limit- 
ing ourselves to aptitudes, but are considering all phases of the man, 
including a variety of traits which come under the broad heading of 
personality and character. 

3 H. M. Johnson, Some neglected principles of aptitude testing, Amcricofi Jour- 
nal of Psychology, 1935, 47, 159-165. 


Of course there are some arguments which could be advanced to 
justify the procedure of totaUng points or scores based upon several 
distinct aptitudes, or even several disparate traits like those included 
in a merit rating-form. First, it might be asserted that if a particular 
trait is so important that the worker is no good without a high rating 
in that trait, the trait will be weighted in proportion to its importance. 
Thus, Johnson would not be able to get an extremely high score as 
an operatic tenor because ability to produce a high note would be 
weighted very heavily in the total score. 

Unfortunately, there may be a number of traits, all equally impor- 
tant, and each of such nature that a low rating would indicate almost 
certain failure on the job. No system of additive point scores could 
take this situation into account. When we are dealing with aptitude 
tests, the process of validation would be such that these possibilities 
would be examined. If there were minimum requirements on each 
trait, these could be established through an examination of the valida- 
tion data which would be secured in developing the test. It would be 
possible to discover critical scores which would be taken into account 
before any composite score is determined. These critical scores would 
appear in the correlation tables which are used in analyzing the valida- 
tion data. If an additive procedure is finally adopted for combining 
the test scores, it can be based upon observed correlations with job 
performance, and the weighting based upon the closeness of relation 
between test performance and job performance. If the additive pro- 
cedure is successful, it must mean that some degree of substitution of 
traits is possible — that the worker can make up for his deficiencies in 
one aptitude by having an extra amount of another. 

These devices for overcoming the difficulty do not exist when we 
are dealing with merit rating. There is no way of weighting the 
various traits according to their importance, except by taking the con- 
sensus of opinion of those engaged in developing the rating scale. 
There is no way of validating these weights by a comparison with 
some independent measure, as there is when aptitude tests are properly 
validated. Since there is no independent measure of the value of the 
worker, there is no way of determining whether a low rating on a 
given trait is indicative of certain failure in the job or in other jobs 
to which the man might be promoted. 

Thus, when we are discussing merit rating, it may be claimed that 
the system of weights attached to the various traits takes their relative 
importance into account. It may also be claimed that a worker can 
compensate to a certain extent for deficiencies in one trait by his 
higher degrees of other traits. These are only claims, however, un- 



[Ch. II 

substantiated by any concrete evidence. In fact it is not possible to 
test these claims by any means known to the writer. 

The logical difficulty which has been discussed here at some length 
seems to be the most fundamental difficulty with merit systems based 
upon a point system. If the first four of the advantages we have 
listed as possible attainments of a carefully designed rating procedure 
are secured in any degree, we should abandon the attempt to sum- 
marize the results in a numerical form. The only case in which the 
numerical summary could be justified would be that in which the 
raters do not actually rate distinct traits. Discussions of the ''halo 
effect" have pointed out the tendency of raters to rate men similarly 
on all traits. If the halo effect is extremely strong, we might find 
that the rater does not really rate distinct traits at all. His separate 
ratings may be no more than repeated ratings of the worker on a 
single scale — a scale of the supervisor's belief concerning the worker's 
value to the organization. 

If this were the situation, a numerical total score might be justified 
as being somewhat more reliable than the individual ratings. The 
advantage would be so doubtful, however, that we should seriously 
question the value of the rating technique. We lose most of the 
advantages which might be expected of the graphic rating method if 
the halo effect is strong, and the method no longer has much useful- 
ness. Would it not be better to ask the supervisor to make a single 
careful rating of the over-all value of the man and let it go at that? 

Ewart, Seashore, and Tiffin, in an analysis of a large number of 
ratings made in one industrial concern, found that there was a very 
strong halo effect.* It was so strong that ratings on one trait cor- 
related with ratings on others to the extent of about .70, on the aver- 
age. Tiffin suggests, however, that the result does not necessarily 
hold if the raters are given better training on the use of the rating- 
forms.^ Bolanovich, analyzing the ratings of technical employees, 
found a somewhat lower average correlation.^ Factor analysis 
showed six significant factors, while Tiffin found only three. 

Thus we are faced with two possibilities. If the halo effect is as 
strong as Tiffin found it to be in his statistical study, few of the 
advantages which are claimed for graphic ratings are left. In such 
a case, no elaborate rating system is justified. If, on the other hand, 

* E. Ewart, S. E. Seashore, and J. Tiffin, A factor analysis of an individual 
merit rating scale, Journal of Applied Psychology, 1941, 25, 481-486. 

5 J. Tiffin, op. cit., p. 245. 

6 D. J. Bolanovich, Statistical analysis of an industrial rating chart, Journal 
of Applied Psychology, 1946. 30, 23-31. 

Ch. ii] 



the halo effect is only moderate in size, as it may be under careful 
training of raters, we may secure the advantages of a more complete 
analysis of the individual, and possibly some of the other benefits 
listed above under headings 1-4. We cannot in this case, however, 
justify the use of the total point score to summarize the results. 

If the reader is still inclined to believe that substitution of traits 
is useful as a working hypothesis, in spite of the preceding discussion, 
we should like to point out other difficulties that still appear after this 
assumption is made. Another aim of the graphic rating method is to 
help the raters to adopt common standards of judgment. If this were 
fully successful, a man in one department who is rated at the "aver- 
age" level in a certain trait would be equivalent to a man in another 
department rated at the same level. Students of rating data do not 
believe, however, that this condition is met in practice, since they 
usually provide methods of "correcting" for the raters' differences 
in standards. 

We shall not here discuss the details of these techniques of correct- 
ing for constant errors of the rater. They may be found in numerous 
presentations of rating methods."^ These methods, regardless of their 
details, all have a common basis. If one rater differs from another 
in the average ratings for all men in the department, it is assumed 
that this difference must be an error due to differences in standard 
between the supervisors. Before men from the two departments are 
compared, their ratings must be corrected for this constant difference 
between the raters. To put it another way, these methods assume 
that if the raters differ it must be due to a difference in leniency or 
kindheartedness. The possibility that the men in one department are 
actually better than those in the second department is not taken into 

It is equally unjustified to make the opposite assumption that the 
difference of raters is due to a real difference in the quality of their 
men. No one knows which is the case, and the facts, if they could 
be found, would probably indicate that part of the difference is due 
to difference in leniency of the raters, and part is due to real differ- 
ences between men in the two departments. Either the "corrected" or 
the "uncorrected" ratings, or both, may be in error by unknown 
amounts. What we have said here concerning differences in average 
scores is equally true of differences between raters in the spread of 
their ratings — another difference which is usually taken into account 
in "correcting" the ratings. 

7 For example, J. P. Guilford, Psychometric Methods, New York, McGraw-Hill 
Book Co., Inc., 1936, p. 274. 



[Ch. II 

Thus, even if we are willing to make the assumptions necessary for 
using point scores for rating, there is little chance that we could with 
any degree of confidence compare ratings which are given by different 
supervisors. It is even questionable whether men rated by the same 
supervisor could be accurately compared if they happened to be work- 
ing upon different jobs. The same rater's leniency may vary with 
the type of work, and we should be faced with the same dilemma we 
met in comparing results of different raters. 

So far we have said nothing of the problem of reliability of the 
ratings. The direct study of the repeated ratings of a supervisor 
shows that they often have poor consistency, and the correlations 
between successive ratings by the same rater are moderate at best. As 
Tiffin points out,® the reliabilities are such that a point system of 
rating, based upon a range of four or five hundred points, is absurd 
because differences in the last place and even in the tens place are likely 
to be the result of chance variations rather than real variations 
between the persons rated. This difficulty with reliability could be 
overcome by using only five to seven rough groupings for the total 
score, rather than attempting to use a finer scale, and it might also be 
improved by better training of the raters. Improving the reliability 
of the ratings would not, however, overcome the objections to a point 
system which we have discussed in preceding paragraphs. 

What is the solution of this problem ? As we indicated at the begin- 
ning, there is a real need for intensive search for other modes of 
approach to the problem of evaluating employees. So far, these other 
methods are not available. Until they are, however, the writer be- 
lieves that little is accomplished by elaborating and complicating the 
rating methods. Instead, he would recommend proceeding on a much 
simpler common-sense basis. The following suggestion is presented 
as one possible way out, not as a final solution of the problem. 

A Suggested Procedure of Rating. — For lack of any better and 
more reasonable suggestion, we could assume that experienced super- 
visors ought to be able to distinguish, on an absolute basis, three levels 
of over-all value of men in their departments. If the men are en- 
gaged in quite distinct types of work, they would have to be divided 
according to occupation and rated separately and with reference to 
that occupation. In addition, it seems reasonable that the supervisors 
should be able to detect extreme degrees of certain specialized traits, 
even though they do exhibit the halo effect and unreliability when they 
are asked to rate all men in the department for each trait. 

8 J. Tiffin, op. cit, p. 254. 

Ch. ii] 



The results of these suggestions would be a form like Fig. 27, 
with certain changes dictated by the industry or the occupations in- 
volved in a particular application. The form which is reproduced in 
Fig. 27 is not meant as a final version, and revisions of the instruc- 
tions are probably necessary for use in any specific organization. The 
general approach is the thing which we are mainly interested in pre- 
senting at this time. It should be emphasized that this proposed form 
does not include anything radically new. In fact, it is basically a re- 
turn to procedures of evaluation which are older and simpler than 
current rating techniques. It is an attempt to secure some of the 
advantages of a standardized rating technique without depending so 
heavily upon the assumptions required for the graphic scale-point 
method of evaluation. 



(This should be a single job or group 
of similar, related jobs. Jobs requiring 
unrelated skills or training should be 
considered independently.) 

1. All-round Value 

1. Outstanding Men: List names of the men who are doing an exceptionally 
good job, and whose all-round value to the department is very high. Con- 
sider their present ability and skill, not what you think they might develop 
later. Men placed in this group should be in about the top 10 per cent of 
men in this line of work anywhere, and not only in this shop. If you are 
doubtful about any man's belonging in this group, do not include him. 
List these men in order from top down if that is possible. 

2. Poor Workers: List the names of all men who are doing an exceptionally 
poor job, and whose all-round value to the department is very low. Con- 
sider present ability and skill, but do not include those who are poor 
because they do not have sufficient training or experience. These will be 
listed separately. As in the top group, if you are doubtful about a man's 
belonging in this group, do not include him. The men in this group should 
be those who would be considered as about the bottom 10 per cent of men 
in this occupation anywhere, but they may include more or less than 10 
per cent of your own men. 

3. Average Workers: List all men who have not been included in the first 
two groups, except those who are not fully trained or who lack experience. 
As you list them, consider each one so as to make sure that he does not 
belong either in Group 1 (exceptional men) or Group 2 (poor workers). 

4. Trainees and Inexperienced Workers: Include here all those who have not 
finished the normal training period, or who have not worked long enough 
to show what they can do. Mark those who show promise with a G, and 
those who do not appear to make normal progress with a P. 

Figure 27a. Suggested Form of Employee Appraisal (page 1) 



[Ch. II 

II. Special Qualifications and Abilities 

Listed at the left are a number of special characteristics which may be im- 
portant considerations in making promotions or transfers. List, for each of 
these quahfications, the names of those men in the job you are considering 
who are outstanding in this quaHty, and those who are especially poor in 
this respect. The names listed are not necessarily the same names you have 
given in your over-all estimates of the value of the employee on page 1. 

Quality * Outstanding Men Poor Workers 

1. Dependability. (Does he 
follow instructions with- 
out close follow-up?) 

2. Ability to get along with 
others. (Is he respected 
and trusted by fellow 

3. Specialized technical 
knowledge of the job. 

4. Ability to instruct others 
in his job. 

5. Interest in his work and 
in the department. 

* Other traits to be added to meet the needs of a particular organization. 
Instructions to be revised after conferences with supervisors and experience 
in application of the forms. 

Figure 27b. Inventory of Personnel (page 2) 

The writer believes that this 'Tnventory of Personnel," though 
still having flaws which cannot be entirely eliminated, is more defen-' 
sible than the graphic rating scale. It obtains an over-all estimate of 
the value of men in a job from the supervisor, which, in view of the 
halo effect, is all we can confidently expect of the graphic rating tech- 
nique. It requires the supervisor to make only broad groupings, 
which, in the light of the known reliabilities of the graphic scale, are 
the only groupings justified. (If three groups are felt to be too 
broad, the same technique could be used with four or five categories 
instead of the three major groups illustrated in Figure 27.) The 
method obtains these broad groupings without resort to dubious point- 
rating scales. 

The method also allows us to obtain useful information upon special 

Ch. ii] 



traits without forcing the supervisor to rate all workers on each trait. 
He can thus designate those who show noticeable excellence or defi- 
ciency, without being asked to make ratings on these traits where no 
information is available. 

Last, but by no means least, this procedure should reduce the 
amount of ''paper work." Though the supervisor has to write names 
instead of making check marks, he is not overwhelmed with a rating- 
form for each man. One form can serve for all the men on a job or 
a related group of jobs, so long as sufficient space is provided for list- 
ing the names. At the same time, the supervisor is required to con- 
sider each man because the form requires that each man must be listed 

Obviously, the inventory method here described cannot fit all re- 
quirements any more than can any other single procedure. It is felt, 
however, that it has advantages, both logical and practical, over the 
graphic rating method for a great many of the purposes for which 
the graphic method has been employed. 

If it is still felt that the graphic form of recording supervisor's 
judgments is desirable, some of its difficulties could be avoided by 
other methods of handling the information it contains. The rating- 
forms might be sorted by inspection into three or four categories. 
With some training for those who do the sorting, it is possible that 
the method could be made quite reliable, and it would not be necessary 
to make the assumptions required by a point system of evaluation. 
The evaluators could take into account both the average level of the 
ratings for a given individual, and any marked deficiencies on single 
traits. In this way a rough qualitative analysis of the forms might 
produce a more adequate evaluation than would be possible by the 
point system. 

Even this procedure would not, however, settle the question of 
whether to ''correct" for the rater's tendencies to rate higher or 
lower than normal. That problem could never be settled unless there 
is some more objective criterion against which the ratings may be 
checked. If we had such a criterion, the ratings would no longer be 

Conclusions on the Graphic Rating Method. — We have supported 
the thesis that the graphic rating scale with numerical scoring has 
serious logical- difficulties and practical inadequacies. These flaws 
are so serious that they more than offset any advantages which are 
claimed for these rating techniques, especially since there is little more 
than common-sense argument to support these claims. 



[Ch. II 

There is a serious need for research to develop new and funda- 
mentally different methods of evaluating workers for promotion, 
transfer, or validation of employment techniques. Until this develop- 
ment is carried out, the writer has proposed to return to simpler pro- 
cedures which require fewer doubtful assumptions. This simpler 
approach was illustrated by a suggested form for recording informa- 
tion in an "Inventory of Personnel." Such an inventory is aimed at 
securing some of the advantages of currently popular rating tech- 
niques, while making the task of the rater as simple as possible, and 
making fewer baseless assumptions. 

Job Evaluation 

Several points in the above discussion of merit rating apply almost 
without change to the problems of job rating. It is only necessary 
to make the minor changes required by the shift of emphasis to 
the job rather than the man. For those who are unfamiliar with the 
methods of job evaluation, however, it may be well to specify the 
method in a little more detail, and to show the general way in which 
the previous criticisms of merit rating would be rephrased in their 
application to job evaluation. 

The process of job evaluation has, like merit rating, been carried 
out in unsystematic fashion for centuries. A similar trend toward 
increased systematization of the process and increased ''objectivity" 
of method has developed in recent years. Again, there are variations 
in the detailed procedure employed, but the basic problems remain the 
same in the various specific methods. 

The job evaluator must first decide upon a list of job character- 
istics which will furnish the basis for comparing jobs. He may, for 
example, use the list of factors given in Table 25, which shows the 
National Electrical Manufacturers Association's system of factors.^ 

The analyst then decides, upon the basis of judgment, the relative 
importance of these factors in rating the value of a job; (to be more 
exact, the value of the time of an average worker of average effort 
and working under average conditions). A scale of points is then 
assigned to each factor and a job rating scale drawn up in a manner 
similar to that used in merit rating. As a result of these steps, the 
number of points assigned to the factor of education, for example, 
may range from to 40, while experience may be graded from to 

9 C. H. Lawshe, Jr. and G. A. Satter, Studies in job evaluation: I. Factor 
analyses of point ratings for hourly paid jobs in three industrial plants, Journal 
of Applied Psychology, 1944, 28, 189-198. 


100, reflecting the judgment that the experience requirement is two 
and a half times as important as educational requirement in the value 
of a job. 

Factors in Job Evaluation 



Initiative and Ingenuity 


Physical Demand 

Mental or Visual Demand 


For Equipment or Process 
For Material or Product 
For Safety of Others 
For Work of Others 

Job Conditions 

Working Conditions 
Unavoidable Hazards 

Each job is then rated, and a total point score determined for it. 
Jobs may then be grouped into classes or labor grades, and to each 
grade is assigned a range of pay. Men in a given labor grade v^ill 
then be paid upon the basis of this standard rate for the labor grade, 
with a correction dependent upon their merit rating, or bonuses for 
above-standard production. Job evaluation determines, therefore, 
the relative base pay for each job. The absolute amount of pay 
depends also, of course, upon the general economic conditions, the 
amount of profit alloted to labor costs, collective bargaining, and 
so on. 

Let us look briefly at the parallels between problems of accuracy 
of job evaluation and of merit rating. At first glance, the relative 
weighting of job factors might seem to be open to the same doubts 
that were raised for individual traits. The methods of job evaluation 
assume that one factor can be substituted for another, just as merit 
rating assumes that one personal quality can be substituted for an- 
other. In job evaluation, however, there is probably better justifica- 
tion for this assumption. A job requiring no education but great 
risk certainly should be paid on a basis comparable to a job requiring 
much education but no risk. The problem of relative weighting is 
still insoluble except on the basis of opinion, but substitution of one 
trait for the other is conceivable and more readily justified in job 

The halo effect would appear at first to apply only to evaluations 
of a worker, and not to evaluations of a job. Nevertheless, there is 
something quite similar, if not identical, in job evaluation. This 
would be a tendency for a job which requires (for example) long 



[Ch. II 

training to be rated high on other factors as a result of this one re- 
quirement. In fact, one statistical study has shown that the XEMA 
job ratings on specific factors are so closely intercorrelated that all 
of the variation in total point rating could be accounted for in terms 
of two or three independent factors. ^° Indeed, when only three fac- 
tors were selected for rating, the ratings of the jobs were practically 
the same as they were in the more complicated rating based upon 
eleven factors. 

As in the case of merit rating, the procedures are frequently used 
without check upon the reliability of the ratings. There is very little 
information available about the reliability of the method. When wt 
turn to the question of validity, we find that job evaluation, like merit 
rating, cannot be validated. Both are estimates of something which 
cannot be measured and must be accepted on faith. 

Except for a few points of difference, therefore, the parallel be- 
tween the two applications of rating can be very closely drawn. As 
a result, our conclusions upon job evaluation will be very similar to 
those for merit rating. Job evaluation procedures should not be used 
without adequate analysis of the reliability of the results. The fine- 
ness of job grading must be in keeping with the reliability. Since 
validity cannot be established except by discussion of the basic 
assumptions made in the method, this is a matter for debate. Until 
workers are convinced that the various factors are given their proper 
relative weight, the procedures will not aid in providing adequate 
incentive conditions in industry. 

Regardless of how we decide to gain common acceptance of the 
method as being valid, the evidence now available indicates that only 
a few factors should be considered. In fact, the whole procedure 
should be kept as simple as possible. Increasing complication not 
only adds nothing to the accuracy, but also makes the method more 
difficult for the worker to understand and accept. 

10 C. H. Lawshe, Jr. and G. A. Satter, op. cit., and Studies in job evaluation: 
II. The adequacy of abbreviated point ratings for hourly paid jobs in three indus- 
trial plants, Journal of Applied Psychology, 1945, 29, 177-184. 

Chapter 12 


It is obvious that accidents are inefficient, so much so that we 
should be wilHng to expend a great deal of extra effort and money in 
order to reduce the danger of accidents. It is also self-evident that 
many of the factors which increase the danger of accident are psycho- 
logical as well as mechanical, chemical, or electrical. It might be said, 
in fact, that the control of the physical factors in accidents has now 
progressed far beyond our ability to control the psychological causes. 

Although few will need to be convinced that accident control is an 
important problem, most of us have a very inadequate notion of just 
how enormous the problem is. In 1940, for example, there were 
about 96,000 deaths in the United States and about 9,000,000 in- 
juries, all due to accidents. During the recent war, from December 
7, 1941 to the surrender of Japan, our war ca,sualties totaled 261,608 
killed and 651,911 wounded, with 32,811 missing.^ The total acci- 
dents in the United States during that same period resulted in 
355,000 deaths, with 36,000,000 injuries. Table 26 shows how these 
totals are subdivided into various main groups. 

TABLE 26 * 

Total Accidents: December 7, 1941 to August 14, 1945 





Motor vehicle accidents 



Home accidents 



Accidents to workers: 




Occupational accidents 



Off-the-job accidents 



* From National Safety Council, Inc., op. cit. 

General Psychological Problems in Accident Control. — It is, per- 
haps, within the realm of possibility that accident-/^roo/ machines 
and transport equipment could be designed and constructed. The 

1 National Safety Council, Inc., Accident Facts, 1946 edition, p. 17. 




[Ch. 12 

expense would probably be prohibitive, and even then we would not 
eliminate many of the accidents that are due to falls, bumps, and 
burns. Under present conditions, our machinery and equipment are 
less dangerous than they formerly were, and much care has been given 
to improvements in their safety characteristics. In the future, the 
greatest reductions in accident rates would appear to depend upon 
control of the individuals who have the accidents. Naturally, this 
broad problem has many subdivisions which must be considered inde- 
pendently of one another. 

For convenience, we may divide the factors affecting accident rates 
into two main groups — general factors which affect most individuals 
in a similar fashion, and individual factors which make some indi- 
viduals much more susceptible to accidents than others. 

Propaganda and ''educational" campaigns to reduce accidents 
would be included logically under the heading of general factors, 
since these campaigns are broadsides directed at the whole public or 
at workers in general. We shall make only brief comments about this 
propaganda factor, however. For one thing, these campaigns have 
grown up along, with the development of various mechanical safety 
devices, improvements in factory design, methods of work, and all 
the changes in industry which have taken place in the past few 
decades. The separate effects of safety propaganda are therefore 
difficult to evaluate. There is little doubt that these campaigns, and 
the activities of such organizations as the National Safety Council, 
have had their share in the reduction of accident rates which has taken 
place during recent decades. It is the amount of their share which is 
an unknown quantity because of a lack of controlled investigation. 

Propaganda for the reduction of accidents is essentially like com- 
mercial advertising or any other form of publicity. In order to 
improve it and make it still more effective than it is, it would be neces- 
sary to resort to the techniques of marketing research and methods 
of testing advertising effectiveness. Those who are especially inter- 
ested in this problem should therefore look into the large literature 
upon these research techniques.^ 

Other general factors whose effects upon accident rate have been 
investigated include most of the specific factors governing efficiency 
which were discussed in Chapter 5 — lighting, ventilation, hours of 
work, alcohol, and fatigue. Since the main outlines of these topics 
have already been presented, we need add only the special details 
related to their effect upon accident rate. 

2 E.g., see H. C. Link, The New Psychology of Advertising and Selling, New 
York, The MacmiUan Co., 1932. 

Ch. 12] 



1. Fatigue. The use of accident rates as a measure of the cost 
of work was taken up along with other indicators of fatigue and 
effort. This was not meant to imply, however, that we were assuming 
any direct correlation between effort, fatigue, and accident rate. Ac- 
cidents themselves are costly, and as such must be entered in the 
ledger under the heading of cost of work or input. The relationship 
of accidents to other factors in cost of work still remains, however, to 
be discovered. 

There is a common belief that accidents increase when the worker 
or driver is fatigued. There is no doubt that falling asleep is a defi- 
nite and unequivocal cause of accidents. Falling asleep is not, how- 
ever, a necessary result of fatigue, and it may occur as a result of 
other factors as well — we have little more than speculation upon the 
relationships involved. Falling asleep may be the result of bad 
ventilation, poor lighting, alcohol, carbon monoxide gas, bore- 
dom, poor motivation, and so on. Driver-asleep accidents do consti- 
tute a serious problem, and one which awaits further clarification by 
studies of factors in sleep as well as by direct analysis of this special 

The National Safety Council has made some statistical studies of 
the driver-asleep accident, although the collection of accurate statis- 
tics is extremely difficult.^ As would be expected, most of these acci- 
dents occur during the hours between midnight and six in the 
morning. The peak hour for truck accidents occurs earlier than that 
for passenger cars. For passenger cars, the highest rate of acci- 
dents in which the driver is asleep occurs about eighteen or twenty 
hours after the last sleep of the driver. Among truck operators there 
is also a high rate of this kind of accident, about five hours after the 
last sleep. The authors of this study suggest that the truck drivers 
obtained so little sleep that they became drowsy more quickly. 

An estimate of the absolute frequency of accidents involving sleep- 
ing drivers is almost impossible to make because even if the driver 
lives through an accident he is reluctant to admit that he was asleep. 
There are enough substantiated cases, however, to indicate that the 
problem is serious. The National Safety Council Report also in- 
cludes data secured by stopping trucks on the highway and asking 
them when they had last slept and how much. As many as 10 per cent 
of the drivers thus questioned had been on a trip without rest as long 
as twelve hours, and the same proportion had been awake as long as 
eighteen hours.* 

3 National Safety Council, Inc., How Long on the Highway? 1937. 
* Ibid., p. 22. 



[Ch. 12 

The problem of permissible driving hours for vehicle operators 
has been attacked by means of fatigue tests. We have already illus- 
trated this mode ot approach in our general discussion of fatigue tests 
where v^e referred to the example of Ryan and Warner. (See pages 
82-86.) The difficulties of interpreting the results of such a research 
were outlined in some detail on those earlier pages. While the findings 
include significant changes in a variety of functions as a result of 
a long period of driving, there is no way of determining w^hen the 
fatigue is serious enough to make driving hazardous. 

A somewhat similar approach was made by a large group of inves- 
tigators whose work was sponsored by the Interstate Commerce 
Commission.^ This research has been severely criticized by Johnson, 
not only on the grounds which make most fatigue-testing programs 
of doubtful practical value, but also for additional special features of 
the technique employed in this particular program.^ Instead of test- 
ing the same drivers before and after driving, and also on control 
days when they did not drive, the investigators resorted to mass test- 
ing of drivers (usually a single battery of tests for a driver). Then 
the drivers were classified into groups according to the number of 
hours they had driven since their last sleep, and the groups were 
compared for differences in performance on the tests. This procedure 
obviously introduces many additional uncontrolled variables which 
make it difficult to obtain significant results. Differences between 
groups would be affected by the different individuals included in 
them, although this could be controlled by large numbers if there w^as 
no systematic error of selection in the various groups. In addition, 
as Johnson points out, the norm with which each score was compared 
was based upon drivers who had not driven since their last sleep, the 
effect of their other activities not being taken into account. 

Finding that four tests showed significant changes with increasing 
hours of driving, the authors then proceeded to work out a composite 
index of fatigue based upon these tests (speed of tapping, reaction 
co-ordination time, simple reaction time, and hand steadiness). 
Against this procedure Johnson raises the same objections which we 
have discussed earlier in merit ratings, where a composite score is 
worked out for a number of disparate qualities of the worker. Such 
a composite is, according to Johnson, a meaningless number which 

5 B. F. Jones, R. H. Flinn, and E. C. Hammond (together with other collabor- 
ators), Fatigue and hours of service of interstate truck drivers, U. S. Public Health 
Service, Public Health Bulletin No. 265 (1941), xxiii and 286. 

6 H. M. Johnson, Index numerology and measures of impairment, American 
Journal of Psychology, 1943, 56, 551-558. 

Ch. 12] 



contributes nothing to the solution of the problem of permissible 
hours of work for truck drivers. 

Regardless of whether Johnson's criticism of the composite score 
is justified, the value of such a procedure of testing with or without 
a composite score is questionable for the reasons we have outlined 
earlier. It might also be mentioned that those differences in test 
score which were found to be significant were not large in a practical 
sense. They are significant because of the large numbers tested, but 
there is much overlapping between groups. The overlapping may be 
due to the poorly controlled research technique, or it may be that these 
tests are affected in a variable manner by driving, regardless of the 
adequacy of control. 

It should also be said that a variety of medical tests were included 
in the program, mostly with negative results. The blood count of 
white corpuscles showed a significant change, with the highest value 
for the group who had driven an intermediate number of hours. 
None of the findings, however, made it possible to set limits to driving 
hours upon medical grounds. 

Johnson suggests that a more useful first step in investigating this 
general problem would be to collect records showing the number of 
hours driven by each person involved in an accident. With proper 
statistical treatment, and with adequate numbers in the samples in- 
volved, it should be possible to determine the effect of length of work 
upon liability to accident. Since this is, after all, the main problem 
in accident control, it would appear more sensible and more direct 
to make this approach first. 

Fatigue and Industrial Accidents. In seeking evidence as to the 
influence of fatigue upon industrial accident rate, most investigators 
have resorted to a study of hourly accident rates. If accidents in- 
crease in frequency as the day progresses, the assumption would be 
that fatigue is responsible for the accidents. A rough hourly pattern 
of this kind has been frequently observed (Figure 28), and the as- 
sumption of the influence of fatigue has followed the discovery. 

Vernon, however, has raised strong doubt as to the validity of this 
assumption."^ In several extensive surveys in munitions plants during 
the first World War he was able to find no marked increase in the 
accident rate in the afternoon as compared with the morning rate. 
(See Table 27.) He points out that variations which do occur from 

From H. M. Vernon, Accidents and Their Prevention. By permission of 
Cambridge University Press (England) and The Macmillan Company, New York. 



[Ch. 12 

hour to hour are closely related to the rate of work. The peak acci- 
dent rate came in the next to the last hour of the morning, which was 
also the period of highest hourly production. This was the pattern 
for day shifts. In night shifts the pattern tended to be much differ- 

1 100 I 1 1 1 r— I 1 « ' 1 1 > 1 

Time of day 

Figure 28. Hourly Incidence of Accidents in Various Industries 
(From Vernon, op. cit., p. 66. By permission of The Macmillan Co., New York.) 

ent, with the maximum accident rate at the beginning of the shift and 
a progressive decline from that time throughout the working hours. 
At night, therefore, the factors differ from those that obtain during 
the daytime shifts. Vernon believes that the difference lies in the 
fact that the night-shift workers have been active with outside affairs 
before coming to work, thus tending to be distracted by these affairs 


and consequently in a lively state. During the day shifts, ''liveliness" 
tended to follow the same pattern as the output rate. 

TABLE 27 * 

Comparison of Morning and Afternoon Accidents 

Relationship of 
Afternoon to 
Morning Accidents 

10 industries running 57.1 

hours per week 96 to 100 

8 industries running 49.9 

hours per week 108 to 100 

Mean: 18 industries running 53.9 

hours per week 101 to 100 

* Vernon, op. cit., p. 67. 

Vernon did find, however, some evidence of an influence of fa- 
tigue. When the working day was exceedingly long (twelve hours), 
the accident rate for women was much higher on an hourly basis than 
it was for a ten-hour day. It was also noted that the afternoon acci- 
dent rate for women was 45 per cent higher than the morning rate 
in the twelve-hour day. (See Table 28.) Increased length of the 
working day did not have the same effect upon men, however. Their 
accident rate remained about the same on an hourly basis, and there 
was no increase in the ratio of morning mishaps to afternoon ac- 

TABLE 28 f 

Effect of Length of Working Day Upon Increase of Accidents 
IN the Afternoon 

Length of 
Work Day 


Total Cuts 

.2 hours 

10 hours 


November, 1915 to 660 
January 1916 

February, 1916 to 2304 
March, 1917 



„ , ^. . . Percentage of 

Relative Acci- ..^ a -j ^ 

, , Afternoon Accidents 

dent Frequency ^ , ^ 

cZ-c. as Compared to 

per Shift TV , . A -J i 
Morning Accidents 









\ Ibid., p. 68. 

The greater susceptibility of women workers to long hours was 
also reflected in the records of the first-aid room in the factories under 



[Ch. 12 

study. There was a marked increase in cases of faintness in the 
twelve-hour day, as compared with the ten-hour period. Vernon re- 
ports this change on a relative basis — the ratio of women's cases to 
those of the men changing from 3 to 1 to 9 to 1. 

Vernon's careful analysis would therefore indicate that progressive 
fatigue during the working day is not an important factor in acci- 
dents for our present industrial labor standards. At least the fatigue 
factor is not clearly evident in the over-all statistics of a whole fac- 
tory or industry. If statistics for certain dangerous and exacting 
occupations could be collected in sufficient numbers, fatigue might be 
found to have a more marked and serious effect. In addition, it is 
possible that the effects of fatigue are disguised by the more powerful 
factors of rate of work and what Vernon calls ''liveliness." If these 
factors could be studied separately and allowed for in the analysis of 
fatigue as a factor, our conclusions could be more definite. 

After Vernon first made this analysis, the United States Public 
Health Service made an analysis of accidents in which the effect of 
production rate was held constant.^ The method was to divide the 
accident index by the production index for a given period; in other 
words, to report in terms of accidents per unit of output. In the 
earlier hours of the day, this accident index rises and falls with the 
output rate. (See Figure 29.) That is, production increases bring 
with them a disproportionate increase in accidents, as noted by Ver- 
non. In the closing hours of a ten-hour day, however, this relation- 
ship breaks down, and the accident index remains high relative to 
production rate. Thus, by more detailed analysis it is possible to 
show an effect of fatigue upon accident rate, once the more potent 
factor of production rate is controlled. 

2. Alcohol. The effect of drinking upon traffic accidents is 
difficult to isolate, since the driver who has imbibed is likely to be 
suffering from lack of sleep, unusual meals, or the excitement of a 
party or date. The tabulation of accidents in which drivers who had 
been drinking were involved will not indicate the importance of 
alcohol per se, therefore, but the total result of all these factors asso- 
ciated with drinking. This is our only source of information upon 
the problem, however, since controlled experimentation is obviously 

8 J. Goldmark, M. D. Hopkins, P. S. Florence, and F. S. Lee, Studies in indus- 
trial physiology: Fatigue in relatipn to working capacity, 1. Comparison of an 
eight-hour plant and a ten-hour plant, U. S. Public Health Service, Public Health 
Bulletin No. 106, 1920. Although this paper appeared soon after Vernon first pub- 
lished his analyses, the latter takes little account of these findings in his more 
recent book, already cited. 

Ch. 12] 



out of the question. Laboratory studies are sometimes made the 
basis of conclusions concerning the dangers of driving when under 
the influence of alcohol, but we have already pointed out that the 
laboratory tests do not necessarily correlate with the competence of a 
subject who is performing a different task in operating a machine or 

Figure 29. Accidents and Production in Eight-hour and Ten-hour Metalworking 


Eight-hour plant data include one year for three shifts, ten-hour plant data, two years 
for one shift. 

(From Goldmark, et. al., op. cit., p. 115.) 

The central problem concerns those who have taken relatively 
small amounts of alcohol that are insufficient for definite intoxica- 
tion. Everyone would recognize the danger of allowing an inebriated 
person to drive. The problem is how much alcohol will produce a 
significant increase in liability to accidents. 

From the scanty data which are available, it would appear that 
even small amounts of alcohol increase the average accident rate. 
Vernon reports experiments in which driving skill was tested in an 
apparatus which closely resembled the actual performance of driv- 
ing.^ After imbibing two or three ounces of whisky, the subjects 

9 H. M. Vernon, op. cit., pp. 159, 166. 



[Ch. 12 

increased their speed of driving but decreased their accuracy of con- 
trol. Usually they were not aware that they had speeded up. More 
direct evidence of the influence of alcohol upon accident liability is 
afforded by the data of Holcomb.^° He tested the alcoholic content 
of the urine of drivers injured in accidents. The percentage of 
these drivers whose tests showed that they had been drinking was 
several times as great as that among a random group who had not 
been involved in accidents. The proportion of injured drivers whose 
tests were indicative of intoxicating amounts of alcohol was thirty- 
three times as great as that of the non-accident group. 

There is still less direct evidence of the effect of alcohol in indus- 
trial accidents. Vernon cites over-all reductions in accident rates 
which followed wartime restrictions on alcoholic consumption, and 
hours of work which prevented drinking. Such correlations must 
be treated with caution, however, since there are so many other fac- 
tors which might make them spurious. 

As Johnson has pointed out, the relationship of accident liability 
to alcoholic consumption is likely to be very complex. Johnson was 
referring especially to the tendency toward compensation. It is even 
possible that a driver could improve his driving performance after 
drinking, because he knows the danger and devotes himself to his 
driving with extra care and attention. Jellinek and McFarland have 
concluded that the effects of alcohol are smaller for habitual tasks 
than they are for the kind of task used in the laboratory or for any 
unpracticed activity. This possibility makes it all the more difficult 
to apply the laboratory results of alcohol experiments to road condi- 
tions. These authors refer to an experiment of Bauer's in which 
real driving reactions were tested after alcoholic dosage. Five out 
of the eight subjects showed a decreased reaction time in this test. If 
we accept Jellinek and McFarland's conclusion, the usual driving 
reactions might not be much affected by moderate consumption of 
alcohol. At the same time, however, we must consider the possibility 
that many situations which produce accidents involve a sudden emer- 
gency that calls for an unusual and relatively unpracticed response. 

The only way in which such questions can be settled, therefore, 
is by direct study of the relations between the ingestion of alcohol 
and actual accidents, with suitable controls to eliminate the influence 

10 R. L. Holcomb, Alcohol in relation to traffic accidents, Journal of the Ameri- 
can Medical Association, 1938, 111, 1076-1085. 

11 H. M. Johnson, Pre-experimental assumptions as determiners of experimental 
results, Psychological Review, 1940, 47, 338-346. 

12 E. M. Jellinek and R. A. McFarland, Analysis of psychological experiments 
on the effects of alcohol, Quarterly Journal of Studies of Alcohol, 1940, 1, Z72-371. 

Ch. 12] 



of other independent factors. What data we have indicate a steady 
increase in liability to driving accidents correlated with increases in 
the alcoholic content of the blood. Whether the same relationship 
holds for industrial accidents remains to be determined. 

3. Lighting and Temperature. The effect of poor visibility 
upon accident rates is so obvious that it requires little analysis. This 
does not mean, however, that the problem is simple. There is still 



47.50 52.50 57.50 62-5 

72.50 77.50 

Figure 30. Accident Frequency in Relation to Temperature (Shell Factory) 
(From Vernon, op. cit., p. 76.) 

much to be learned about highway lighting, headlights, visibility of 
signs and pedestrians, adequate lighting of machinery, and so on. 
Our previous discussion of the general problems of efficient lighting 
have shown the complexities involved. To detail the problems of 
vision in relation to accidents in industry and on the highway would 
require a separate treatise of a size out of proportion to the factual 
information available. 

Temperature has an effect by increasing the hazards of the road, 
but it also has an effect upon the worker himself which sometimes 
increases his accident liability. Variation in temperature appears to 
affect the rate of minor accidents more markedly than the rate of 



severe accidents. Vernon and his collaborators found that the inci- 
dence of cuts in a shell factory was at a minimum for a temperature 
of 68° to 70°, with increases in frequency as the temperature fell 
below or rose above this range as shown in Figure 30. In coal 
mines, high temperatures increased the over-all rate of accidents and 
the amount of lost time, but the change was most noticeable in those 
accidents which resulted in less than ten days absence from work.^* 

PT 1 1 1 1 1 1 1 1 I 

60"* 65** 70'' 75° 80° 60° 63° 70°. 73° 80° 

Figure 31. Accident Disability in Relation to Temperature (Coal Mining) 
(From Vernon, op. cit., p. 80.) 

We are still unable to explain these relationships. The lower 
temperatures may produce clumsiness and stiffness which would ac- 
count for the increase in minor injuries. The effect of the higher 
temperatures is more difficult to account for. It might be suggested 
that the high temperatures serve as distractions, but it is also possible 
that the change in temperature produces some change in ease of co- 
ordination, or increases fatigue. 

Individual Differences in Liability to Accident 

Although it is commonly recognized that some persons tend to 
have more accidents than others, the full implications of this fact are 
not always realized. Part of the difference in accident liability is 
due to conditions which are out of the control of the individual 

13 E. E. Osborne and H. M. Vernon, Two contributions to the study of accident 
causation : A. The influence of temperature and other conditions on the frequency 
of industrial accidents, Industrial Fatigue Research Board (Great Britain), 1922, 
Report No. 19, 1-17. H. M. Vernon, T. Bedford, and C. G. Warner, A study of 
absenteeism in a group of ten collieries, Industrial Fatigue Research Board (Great 
Britain), 1928, Report No. 51. 

1* Vernon, op. cit. 

Ch. 12] 



worker or driver — different jobs involve different amounts of risk, 
and some automobile drivers expose themselves to greater risks be- 
cause they must drive farther than others or under poor driving con- 
ditions. Table 29, containing data collected by Tiffin in a steel mill, 
shows a wide variation in frequency of mishaps which occur in dif- 
ferent jobs. Information of this kind shows the safety engineer 
where his more serious problems lie. For the psychologist who 
wishes to study individual differences in tendency toward accidents, 
differences in exposure and job hazard represent variables which 
must be held constant before the effects of the individual's charac- 
teristics can be observed. 

TABLE 29 * 

Variation in the Average Number of Hospital Visits per Year 
Among Employees on Eleven Different Jobs 

Average Number of Hospital 
Job Visits per Year 

Craneman 3.55 

Opener 3.54 

Reckoner 2.96 

Machinist Helper 2.77 

Leader 2.75 

Sheet Inspector 2.54 

Shear Helper 2.40 

Assorter 2.36 

Potman 2.10 

Foreman 1.16 

Roll Turner 47 

* From J. Tiffin, Industrial Psychology, New York, Prentice-Hall, Inc., 1942, p. 288. 

Even within a group of workers who have equal hazards in their 
jobs, there will be a consistent tendency for certain workers to suffer 
more accidents than others. Those workers who have a high accident 
rate in one year will also have a high average rate in succeeding 
years. Such a worker, with a permanently high liability to accident, 
is spoken of as accident-prone. 

The psychological problem is to find the causes of such accident- 
proneness. It is insufficient to attribute the accident tendency to 
''carelessness," "clumsiness," or "faulty attitude," as has sometimes 
been done. These terms mean very little, and do not solve the prob- 
lem. We still need to know why the individual is careless or has 
faulty attitudes. 



[Ch. 12 

There has been a considerable number of exploratory researches 
aimed at finding underlying factors responsible for accident-prone- 
ness. The basic procedure of these researches has been very simple. 
A group of accident-repeaters and a group of accident-free workers 
or drivers are given a variety of psychological tests, such as reaction 
time, judgment of speed of movement and distance, muscular 
coordination, and so on. The average score of each group on 
each test is computed in order to see if the repeater group is 
significantly poorer at the test than the accident-free group. 

There are several serious difficulties with such a gross attack upon 
the problem. First of all, the group of SLCc'idtnt-repeafers is probably 
not made up entirely of Siccidtnt- prone individuals. Notice that the 
definition of accident-proneness is in terms of liability to accident, 
i.e., in terms of the probability that the individual will have an acci- 
dent within a certain period of time.^^ It is not stated in terms of 
the actual accident record of the individual. A person who has a 
large number of accidents is not necessarily accident-prone. He is 
likely to be accident-prone, but he may be, instead, a person who has 
had bad luck — who has suffered accidents in spite of a low probability 
of accident. We can illustrate what is meant here by a reference to 
games of chance. If I toss a coin five times and obtain five heads, 
the coin is not necessarily loaded. Even if I repeat the trials and 
again obtain five heads it would not be conclusive proof of loading. 
I might strongly suspect loading, but it is possible to obtain this 
result by chance with a fair coin. Most of the repeater group is 
probably accident-prone if the group is properly selected. Some, 
however, are not accident-prone, and we have no way of telling in 
advance which individuals these are.^® 

15 To be more exact, liability is defined as a limit. It is the average number of 
accidents per year (month, etc.) which an individual would have if he were exposed 
to the same conditions for an infinitely long period of time. Defined in this way, 
liability of an individual can never be measured, but only estimated. It is like the 
"true" score of an individual on an aptitude test, the score which he would attain 
if we had a perfectly reliable test. Accident rate over a period of a few years 
is, however, less reliable as an estimate of the true score, than most aptitude tests. 

In spite of this difficulty we need the concept of accident liability. Accident rate 
cannot be predicted as such because it is determined by many truly accidental factors 
(luck or chance) as well as by permanent tendencies of the individual, and only 
these permanent tendencies are predictable. 

16 Essentially the same point of view is expressed by Cobb in relation to cor- 
relations between test scores and accident records. P. W. Cobb, The limits of use- 
fulness of accident rate as a measure of accident-proneness. Journal of Applied 
Psychology, 1940, 24, 154-159. For a more complete mathematical treatment of 
the problems of accident distribution, see E. M. Newbold, Practical applications 
of the statistics of repeated events, particularly to industrial accidents, Journal of 
the Royal Statistical Society, 1927, 90, 487-547. 

Ch. 12] 



Similarly, the comparison group of non-accident individuals 
probably contains some accident-prone individuals. An accident- 
prone individual may have very good luck and thus avoid accidents 
in spite of his high liability. The result is that v^e are comparing the 
test results of two mixed groups. A difference w^hich might exist 
between truly accident-prone individuals and persons who are truly 
low in accident liability may therefore be partially covered up or re- 
duced by the mixture of the two groups. This is especially likely 
when the accident histories of the experimental groups are known 
over only a relatively short time. In the long run, liability and acci- 
dent record tend to agree, and therefore we must have a long-term 
history of the individual, and he must be exposed to similar hazards 
over that long history if we are to get reliable results. 

In order to make the above ideas more concrete. Table 30 shows 
an example of the distribution of accidents among a large group of 
automobile drivers. These data show the characteristics which have 
been found in a number of similar statistical investigations. The 
column headed ''Actual Number" shows the number of drivers who 
experience 0, 1,2, and so on accidents during the period of the sur- 
vey. The column labeled ''Expected Number" represents the distri- 
bution which would be expected if all the drivers had the same 
liability to accident (a liability equal to the average liability of the 
actual group). The difference between the "expected" and the actual 
distributions indicates that liability must differ from person to per- 
son, as we have already indicated. It also illustrates the point we 
have made with respect to the procedure of validating tests for 

In order that the statistical relations may be brought out, an addi- 
tional distribution of accident frequencies is given in Table 31. This 
distribution is a hypothetical set of data chosen for ease of computa- 
tion, but it will be seen that the general shape of the distribution is 
essentially the same as that given in the real distribution of Table 30. 
This hypothetical distribution was obtained by assuming that there 
were two separate groups of drivers or workers. One group of a 
hundred individuals averages two accidents per person over a period 
of observation. The larger group of a thousand individuals averages 
only 0.5 accidents per individual — only one fourth the rate of the 
accident-prone group. Adding the two distributions together re- 
sulted in the distribution which is labeled the "observed" distribution. 
The "expected" column is based upon the assumption of equal lia- 
bility for the whole group of eleven hundred individuals who have an 
average of .64 accidents per individual. The divergences between the 



[Ch. 12 

"expected" and "observed" values are seen to be like those in the 
actual case of Table 30. 

TABLE 30 * 

Distribution of Automobile Accidents in a Random Sample 
(Connecticut, 1931-36) 

Operators having these accidents 
Accidents per operator 

during experience Actual Number Expected Number Difference 

23,881 23,234 647 

1 4,503 5,572 — 1,069 

2 936 668 268 

Total (0-2) 29,320 29,474 - 154 

3 160 53 

4 33 4 

5 14 

6 3 

7 1 

Total (3-7) 211 57 154 

Total 29,531 29,531 

♦ U. S. Bureau of Public Roads, Motor Vehicle Traffic Conditions in the United States: 
The Accident-Prone Driver. (House Document No. 462, Part 6, 75th Congress. Third 
Session), 1938 (Washington), p. 19. 

In the last two columns of the table are given the distributions of 
accidents in the two hypothetical groups which make up the total. It 
will be seen that a substantial number of the accident-prone indi- 
viduals have zero to one accident apiece. Also, the low-liability group 
contains some individuals with three or four accidents. If w^e w^ere 
to choose for study those accident ''repeaters" who have three or more 
accidents, we should include thirty-two accident-prone individuals 
and fifteen low-liability individuals. The zero accident group w^ould 
include fourteen individuals whose liability is high and should be 
called ''accident-prone." 

It is obvious that this example oversimplifies the problem. In real 
groups of individuals there are likely to be many different levels of 
accident liability, and the total accident distribution must be con- 
ceived as the sum of all these different groups. The same general 
principle of variation in accident rate for persons of equivalent 
liability will still hold, however, so that we have done no violence to 
the principle in choosing a simpler case for illustration. 

Ch. 12] 



TABLE 31 * 

Hypothetical Analysis of Accident Distribution with Two 
Liability Groups 

Number of 




Number of 


Group * (2 ac- 

Group * (.5 



cidents per 

accident per 









































* The "expected", hypothetical accident-prone and low-liability distributions are esti- 
mated from tables of the Poisson distribution. (H. F. Dodge and H. G. Romig, Single 
sampling and double sampling inspection tables, Bell System Technical Journal, 1941, 20, 
Figure 6, opposite p. 20.) 

There is still another condition which tends to vitiate results which 
are based upon the gross procedure of comparing test results of two 
groups of individuals of different accident histories. This is the 
obvious fact that there are many different causes of accident-prone- 
ness. Even if we are able to sort out a reasonably ''pure" group of 
accident-prone individuals, some of the group may be prone because 
of various ocular deficiencies, some because of poor muscular co- 
ordination, others because of certain faulty habits of machine opera- 
tion which have not been corrected, and so on. Suppose, now, that 
we wish to find out whether a specific factor, such as visual acuity, 
is a factor in accident liability. If we follow the gross procedure 
which we have outlined, we administer a test of visual acuity to our 
two groups of workers or automobile drivers. Suppose, also, that 
there is a certain small group of workers who are accident-prone be- 
cause of defective visual acuity, and for no other reason. This group 
will probably be at most a small fraction of the total group of 
accident-prone individuals. The remainder may show normal or even 
superior visual acuity. The net difference in visual acuity between 
the two groups is therefore likely to be so small that it is scarcely 
detectable, even though visual acuity is (hypothetically) a definite 
cause of accidents in certain individuals. When we consider all the 
factors which might conceivably make an individual liable to accident 
in some complex task, it is evident that no single test could be ex- 



[Ch. 12 

pected to show very large differences between groups selected in terms 
of gross accident records alone. Negative findings for any particular 
test may mean either that the test performance is completely un- 
related to accident liability, or that the factor tested is important in 
such a small group that its effect is hidden.^^ 

It might appear that both the difficulties in validating tests of 
accident-proneness which we have discussed apply only to the particu- 
lar kind of analysis which we have described. That is, they might 
appear to be a result of the process of selecting an accident-repeater 
group and an accident-free group and comparing average test scores 
for the two groups. Essentially the same problems arise, however, 
in any procedure for relating scores on a single test to gross accident 
liability. Liability is never directly measured; it is estimated from 
the accident record of the individual. The only way this estimate 
can be improved is by using longer and longer periods of observation. 
Gross accident liability, in turn, is a function of many variables, so 
that no one test can show extremely high correlation with liability, 
to say nothing of accident history. 

Later we shall make some suggestions for new kinds of analysis 
which should establish more meaningful relations between tested 
characteristics of an individual and his accident tendencies. Before 
we do this, however, it will be necessary to become acquainted in a 
general way with the results now available — with the kinds of test 
which have been employed, and with the character of the statistical 
findings. In view of the logical difficulties we have described, it is 
surprising that so many significant relationships do appear in re- 
searches based upon the methods we have described. 

Some Tests of Accident-Proneness. — The more frequently used 
tests may be grouped roughly as follows : 

1. Tests of reaction time and co-ordination (frequently in the form 
of a miniature of the car controls or machine). 

2. Tests of special visual characteristics (judgment of speed, dis- 
tance, sensitivity to glare, peripheral and central acuity), 

3. Distractibility and reaction in an emergency. 

4. Intelligence. 

1. Reactions and Co-ordination. As selected examples of this 
kind of test we may refer to those used by DeSilva and to those 
of Farmer and Chambers. DeSilva, directing an extensive program 

17 See Farmer's discussion of this probJem in his comments appended to I^Iiss 
Newbold's statistical papers, {op. cit.) 

Ch. 12] 



of research upon automobile drivers, made use of a test consisting of 
"dummy" automobile controls (Figure 32) which the subject ma- 
nipulated while looking at a screen upon which appeared signals and 
an image of a highway scene/^ The subject was tested for speed of 
response to a traffic signal (the time from the signal to the first 

Figure 32. Driver Testing Apparatus with Moving Road Scene 
Used for Harvard Tests of Braking Reaction Time, Steering, and "Vigilance Reaction." 

(Reprinted by permission from Why We Have Automobile Accidents, by H. R. DeSilva. 
Published by John Wiley & Sons, Inc., 1942. P. 63.) 

response in removing the foot from the accelerator, and the time 
elapsed until the brake was applied). The accuracy of steering the 
simulated car was also measured, and another reaction speed test 
was applied while the subject was engaged in steering ("vigilance 

18 H. R. DeSilva and T. W. Forbes, Driver Testing Results, Bureau for Street 
Traffic Research, Cambridge, Massachusetts, 1937. 



[Ch. 12 

It is difficult to present a representative sample of DeSilva's re- 
sults with these tests. The differences and the degrees of significance 
varied from city to city in the various parts of his study. These 
variations are attributed by DeSilva to differences in driving con- 
ditions in the various cities, but we might also suspect some variation 
in the manner in which the test groups were selected. For example, 
the volunteer group which is used for comparison may have been a 
different kind of group in each city. Nevertheless the differences are 
all in the expected direction, even though some are not statistically 

More recently, Fletcher has used DeSilva's tests and some others 
in addition, in a comparison of **good" and ''poor" drivers in Cali- 
fornia.^® The good drivers included drivers of heavy and light 
trucks, and also a class of highway patrolmen, making a total of two 
hundred men in all. The poor drivers had been involved in a single 
accident resulting in a death, or in three or more accidents resulting 
in injury. Table 32 shows the whole series of test results upon these 
two groups of drivers. The first four tests listed would come under 
the heading of the present discussion — tests of reaction time and co- 
ordination. It will be seen that the differences, though small and 
indicating a considerable overlapping of the groups, are nevertheless 
statistically significant. 

There may be some criticism of Fletcher's survey because of the 
specialized group of drivers chosen for the ''good" driver class. Their 
test scores might not be considered representative of the performance 
of private car operators who are low in accident liability, and this 
might account for the more significant results which Fletcher ob- 
tained. Truck operators or others whose profession is driving might 
be a selected group with respect to the abilities tested, without the 
abilities necessarily being related to accident tendency. A deficient 
person might be able to compensate well enough for casual pleasure 
driving, but find driving fatiguing if he were to take it up as a full- 
time occupation. 

Fletcher has considered the possibility that other factors have 
affected the test scores. He found that the groups differed in age and 
driving experience, as well as in test scores. The good drivers were 
younger and* had driven a greater number of miles in the year pre- 
ceding the test.^^ When these factors were controlled by selecting 

19 E. D. Fletcher, Preliminary report on special tests: Part I, Capacity of 
special tests to measure driving ability, State of California, Department of Motor 
Vehicles, pp. 1-12. (Mimeographed publication.) 

20 It should be noted that the mileage record of an individual has two implica- 
tions. It implies greater practice in driving, but it also involves greater exposure 

Ch. 12] 



individuals of comparable age and mileage record, the differences in 
test performance still appeared. The poor drivers had the advantage 
of more years of driving experience, if that means anything for 
driving skill. This factor was also controlled in a special comparison, 
and was not found to affect the general results. 


Fletcher's Comparison of Test Performance of "Good" and 
"Poor" Drivers * 

Test Average Scores 

Simple Braking 
Reaction (1/100") 




Differences f 

Ratio X 


Vigilance Braking 






Simple Steering 

(per cent score) 





Vigilance Steering 
(per cent score) 





Speed Estimation 

(error score) 





Glare Sensitivity 






(percent Left 

failing) Right .... 






(ppr cent Both 

failing) Left 






* Fletcher, op. cit., pp. 2-7. 

t All differences are in favor of the good drivers. 

t Critical ratio of 3.00 or greater indicates probability of less than .002 of obtaining 
the difference by chance. 

Farmer and Chambers used a group of tests which they classed as 
''aesthetokinetic." These included the ''dotting" test described 
earlier in its application to the study of fatigue (page 81), a pur- 

to risk. From the latter point of view, the poor drivers are more accident-prone 
than they at first appear to be, since their accidents, per mile, are still more diver- 
gent from the records of the non-accident drivers. 

21 E. Farmer and E. G. Chambers, A study of personal qualities in accident- 
proneness and proficiency, Industrial Health Research Board (Great Britain), 
1929, Report No. 55. (With F. J. Kirk.) Tests for accident-proneness, ibid., 1933, 
Report No. 68. A study of accident-proneness among motor drivers, ibid., 1939, 
Report No. 84. 



[Ch. 12 

suitmeter test requiring the subject to follow irregular movements of 
a pendulum with a pointer, a two-hand co-ordination test in which the 
movement of a pointer is controlled by two handles to be manipulated 
simultaneously, and some other tests of similar nature. The tests 
have been used for workers in several skilled occupations, among 





Figure 33. Accident Rates of Dockyard Apprentices, Related to Test Scores 
("Aestheto- Kinetic Tests") 

Difference between top and bottom groups is statistically significant. 

(From Farmer, Chambers and Kirk, op. cit., 1933, p. 2L Reproduced by permission 
of the Controller of His Britannic Majesty's Stationery Office.) 

them shipwrights and carpenters, and also for bus operators. Figure 
33 shows the accidents of dockyard apprentices who are grouped into 
quarters according to their scores on the aesthetokinetic tests. There 
were 251 subjects learning the trades of shipwright, electric fitter, and 
engine fitter. The top quarter of the group is significantly lower in 
its accident rate than the bottom quarter. 

When Farmer and Chambers applied these tests to bus operators, 
the differences between the high-scoring and low-scoring drivers were 
not statistically significant. The investigators believe, however, that 
the tests may be useful in combination with other selective devices. 

Ch. 12] 



TABLE 33 * 

Table Showing the Mean Percentage Accident Rate Over the 
Whole Period of Exposure of Groups Selected by Different 
Methods, Taking the Accident Rate of the Whole Observed 
Group as 100 Per Cent 

Accident Rate of the Whole Group . . . 

1. Accident rate of those left after re- 
jecting the high-accident subjects in 
the first year 

2. Accident rate of the top three inter- 
quartile groups of the aesthetokinetic 

3. Combination of Method 1 and Meth- 
od 2 

4. Accident rate of the top inter-qua r- 
tile group only in the aesthetokinetic 

Tights Electric Fitters Engine Fitters 

100 100 100 

87 99 94 

80 88 90 

75 88 79 

61 60 74 

* Farmer and Chambers, op. cit., No. 68, p. 28. Reproduced by permission of the 
Controller of His Britannic Majesty's Stationery Office. 

TABLE 34 f 

Effect on the Subsequent Accident Rate of Removing Drivers 
WITH Many Accidents in Their First Year or Removing 
Drivers with Poor Aesthetokinetic Co-ordination 

of Drivers 

Accident rate of the whole group for 
all years except the first 

1. Accident rate after removing drivers 
with three or more accidents in their 
first year ' 28 

2. Accident rate after removing the 
worst 25 per cent in the tests 23 

3. Combining Method 1 and Method 2 44 

Percentage of 




Percentage of 

t Farmer and Chambers, op. cit., No. 84, p. 35. Reproduced by permission of the 
Controller of His Britannic Majesty's Stationery Office. 

The possibility of selecting w^orkers w^ho w^ill have a lower accident 
rate is suggested by Tables 33 and 34 from the same authors. The 
first of these tables illustrates a method of selection based upon the 
accident rate of an apprentice during his first year of w^ork, along 
w^ith the elimination of those with low scores on the tests. The other 
table, for bus operators, is similar, except that a test of perseveration 



[Ch. 12 

and a test of intelligence are included in the data. It will be seen 
that eliminating those drivers who have a bad record in their first 
year is slightly more effective than selection by tests. 

2. Visual Tests. As the examples listed above would indicate, a 
great variety of visual characteristics have been tested in relation to 
accident susceptibility. It is probable that any visual defect is dan- 
gerous in certain occupations. On the other hand, the defective person 
can learn to avoid situations in which his defect will be troublesome 
or dangerous. It is therefore necessary to make direct investigations 
in order to determine the actual correlations between visual defects 
and accidents. DeSilva cites several sets of data indicating that poor 
visual acuity (inability to distinguish small details) is related to 
accident-proneness.^^ Weiss and Lauer, on the other hand, could 
discover no relationship,^^ although acuity did show some correlation 
to a laboratory driving test. Several studies have indicated that 
drivers who have one defective eye are likely to have intersection 
accidents more frequently than those with balanced vision. Table 
35 shows the results of one such study. 

TABLE 35 * 
Vision and Intersection-Type Accidents 

61 Accidents happened to defective-vision drivers on 

defective-vision side 
10 Accidents happened to defective-vision drivers on 

normal-vision side 
32 Accidents happened to normal-vision drivers 


* Fletcher, op. cit. 

We shall not go into details about the other kinds of visual defect 
which may be studied. Table 32, page 273, illustrates some of 
the tests and differences discovered in the California driver study. 

3. DiSTRACTiBiLiTY AND CARELESSNESS. As a part of a battery 
of tests for selecting taxicab drivers, Snow designed a test of "emo- 
tional stability" which was found to be effective for selecting oper- 
ators who would have good driving records. The subject operated 

22 H. R. DeSilva, Why we have automobile accidents, op. ext., pp. 71, 72. 

2 3 A. P. Weiss and A. R. Lauer, Psychological principles in automotive driv- 
ing, Ohio State Contributions to Psychology, No. 11, 1931, p. 58. 

24 A. J. Snow, Tests for transportation pilots, Journal of Applied Psychology, 
1926, 10, 37-51 ; Tests for chauffeurs, Industrial Psychology, 1926, 1, 30-45. 

Ch. 12] 



a kind of switchboard in which he was required to Hght a series of 
Hghts, in order, by manipulating a compHcated arrangement of 
switches. He was also instructed as to what to do if a short circuit 
occurred. The short circuit, which produced sparks and a shock in 
the subject's hand, occurred a certain number of times in each test. 
With this arrangement it was possible to record the speed of reaction 
in dealing with the short circuit, and also to observe the changes in 
skin resistance which accompany certain portions of the test. The 
former measure of reaction speed was used alone for predicting per- 
formance of the drivers. 





















1-3 4 AND 



Figure 34. Relation of Accidents to Scores on Driver Tests (thirty-five cases) 
(From Wechsler, op. cit., p. 27.) 

Combined with an intelligence test, the test of ''emotional stability'* 
was effective in improving the accident records of operators hired for 
the Yellow Cab Company of Chicago. A group of men who were 
hired was subdivided into those who showed satisfactory performance 
on the tests, and those who were considered unsatisfactory. The 
satisfactory group averaged .2 accidents per man in a period of ten 
weeks, while the unsatisfactory group averaged 1.00 accidents during 
an identical period. 

Unfortunately, we do not have comparable data for private car 
operators or for accident liability of industrial workers. Snow's suc- 
cess with commercial drivers would indicate the need for applying 
tests of this kind in studies of accidents in other fields. 



[Ch. 12 

In another investigation of taxicab drivers, Wechsler employed 
a test involving a series of reactions performed with car controls in 
response to light signals. The test was designed to record not only 
the time of reaction, but also errors in reaction such as pressing the ac- 
celerator rather than the brake. Errors are considered a measure of 
''carelessness." The relation of these scores to accident rate for a 
group of thirty-five cases is shown in Figure 34. The reaction speeds 
showed a non-linear relation to accident record. As the speed in- 
creased up to a certain level the accident rate went down. If the 
reaction speed was especially high, however, the accident rate again 

4. Intelligence Tests. These tests have not been used as 
frequently as other tests in the analysis of accident-prone individuals. 
Some investigators report a correlation between intelligence and ac- 
cident rate. Farmer and Chambers, however, could discover no cor- 
relation of any significance in their studies of dockyard apprentices.^® 
The latter result may be due to a narrow range of intelligence among 
these apprentice groups, or it may be that intelligence is related to 
accidents only in certain occupations. Vernon suggests that we 
should expect accidents involving errors of judgment to be related to 
intelligence, but that the kind of accident suffered by the apprentices 
is more likely to involve skill rather than judgment."' When a 
correlation does appear, it may be spurious, however. The workers 
involved are in many different occupations, of differing risk. Conse- 
quently there may be a correlation of intelligence with the kind of 
work, and only indirectly with the accident rate. Only with hazard 
held constant can a correlation of a test with accident rate be meaning- 
ful. At present there is no good evidence that intelligence is related 
to accident liability in industry or in private car operation, although 
it may be an important measure for certain special groups. 

Predicting Accident Rates by Means of Tests. — Most of the re- 
searches so far described have illustrated the possibility of obtaining 
significant differences in test scores between accident repeaters as 
a group and a group of accident-free individuals. Only the studies 
of Farmer and Chambers have furnished data upon the predictive 
value of the tests in sorting out individuals of high and low accident 
liability. Those studies, it will be remembered, indicated that the ac- 

2 5 D. Wechsler, Tests for taxicab drivers, Journal of Personnel Research, 1926, 
5, 23-30. 

2 6 Farmer, Chambers and Kirk, op. cit., 1933, p. 30. 
2 7 H. M. Vernon, op. cit., p. 44. 

Ch. 12] 



cident record of an individual for a year is of more value than the 
tests in predicting subsequent accidents. Other programs, like that 
of DeSilva, have used many more tests and more elaborate pro- 
cedures. We should therefore consider the possible predictive value 
of these other programs, the components of which have been illus- 
trated in the preceding pages. 

Cobb, in a research carried out for the Highway Research Board, 
collaborated with DeSilva and with Lauer in an extensive analysis of 
the predictive value of the Harvard tests (DeSilva) and the Iowa tests 
(Lauer ).^® Some 3600 individuals took the tests, 2800 being given 
both complete batteries. In addition, information from the State 
Bureau of Motor Vehicles, and answers to a questionnaire or inter- 
view, were secured. Each item of information and each test was 
correlated with the accident history as given in the Motor Vehicle 
Bureau records. 

TABLE 36 * 
Correlations with Accident Rate 

Predictive Factor 

Miles Driven per Year + .199 

Education — .190 

Aptitude (intelligence) — .195 

Years Licensure 4- .127 

Drivometer Errors + .112 

(a miniature test from the Iowa Battery) 

* After Cobb, op. cit. 

The highest correlations between individual items and accident 
history are shown in Table 36. Although there were thirty- 
three significant correlations obtained (for seventy-two tests in all), 
all those not mentioned in Table 36 were less than .10. That is, they 
were significant because of the large number of cases involved, but 
indicated little predictive value. 

2 8 P. W. Cobb, Report to the Highway Research Board on the Automobile 
Driver Tests Administered to 3663 Persons in Connecticut, 1936-1937, and the 
Relation of the Test Scores to the Accidents Sustained. (Unpublished manuscript 
in microfilm filed with the Highway Research Board of the National Research 
Council, Washington, D. C, 1939.) Also summarized in H. M. Johnson, The de- 
tection and treatment of accident-prone drivers. Psychological Bulletin, 1946, 43, 

2 9 Cobb applies tests of significance which are strictly applicable only to correla- 
tions involving normal distributions, even though the accident distributions are 
known to be non-normal. We shall come back to this problem a little later. 



[Ch. 12 

There are two reservations to be kept in view in interpreting these 
individual correlation coefficients. First, a perfect test could not give 
a correlation of more than .454. This was computed by Cobb as the 
upper limit because of the unreliability of accident rate as an indi- 
cator of liability. Secondly, Cobb found that the state records of 
accidents were very incomplete, many of the operators having acci- 
dents outside of Connecticut which were not recorded in the Connec- 
ticut files. Thirdly, the group of individuals tested, though large, 
was also very heterogeneous, including a wide range of driving ex- 
perience, of yearly mileages, of types of car operated, and so on. 
These are all factors which would tend to reduce the correlation be- 
tween any test and the criterion of accident rate. 

By means of a combination of predictive factors, Cobb was able 
to form a multiple predictive equation which gave a multiple correla- 
tion of .35, using all the tests (twenty-two) which would contribute 
anything to the prediction. He was also able to obtain a multiple 
correlation of .30 by a combination of factors which did not make 
use of any apparatus tests. 

It is difficult to interpret these results, however, because Cobb has 
included among his predictive factors items like miles driven per year, 
years of licensure, and years of driving experience. The yearly mile- 
age is a factor increasing the risk of accident, but the predictive value 
of tests would be most clearly demonstrated if the risk were held 
constant, allowing the individual differences in liability to appear in 
isolation from the external hazard. The other factors would be re- 
lated to age, and also to possible changes in driving conditions over 
the years. Age and changing conditions should be kept as a problem 
distinct from that of determining the personal defects which con- 
tribute to accident-proneness. 

Cobb also examined the following relationships : curvilinear cor- 
relations and critical scores on the tests ; the possibility of ''sifting" 
a group who were clearly deficient on several tests ; and, finally, the 
correlations of specific tests with specific types of accident. These 
results were largely negative. Although a group could be ''sifted" 
so that it consisted of those with a pattern of several deficiencies, this 
procedure showed less predictive value than the multiple equation de- 
scribed above. Almost no significant relationships between particu- 
lar tests and specific types of accident were discovered. In the latter 
case, the failure may be due to the fact that Cobb examined only 
those possibilities which seemed reasonable a priori, for example the 
relationship of ability to judge distance to front-end and rear-end 
collisions. Perhaps the relationships to be sought cannot be worked 

Ch. 12] 



out a priori. This is made plausible by the fact that some of the re- 
lationships which he examined turned out to be the opposite of what 
had been expected. 

Although Cobb's undertaking was on a large scale, it still leaves 
us with open questions. The results would, perhaps, have been more 
enlightening if he had used smaller, more carefully selected and con- 
trolled groups, rather than the large heterogeneous population studied. 
There is no essential contradiction between his results and those of 
DeSilva and others, since Cobb, like the others, obtained significant 
though small correlations. Cobb's results do indicate, however, that 
the further development and careful validation of driver tests is 

The Uses of Tests for Accident-Proneness. — The preceding ex- 
amples have given a few tests which are representative of a consider- 
able number which have been used at one time or another. In 
illustrating the results of individual tests, we have shown that some 
of the tests have demonstrated value in selecting commercial vehicle 
operators and possibly for selecting workers in mechanical trades. 
Tests of this type have not been used for selection of private vehicle 
operators for several reasons. The most important reason is that 
the tests do not have sufficient validity for such widespread applica- 
tion. To prevent the owner of a car from operating it, the tests 
would have to have a very high predictive value before they could 
gain acceptance from the public at large. In commercial selection we 
may be willing to risk eliminating a few operators who might have 
been perfectly satisfactory because, in doing so, we are eliminating 
many whose liability is definitely high. The only criterion of success 
of the tests is the accident rate of those who are employed, and we 
can afford to run the risk of injustice to some of those who are re- 
fused employment. Obviously this procedure would not be acceptable 
for private vehicle operators. 

Practical applications of tests like those described above to private 
vehicle operators, and in some instances to commercial employees as 
well, have been in the form of ''clinical" treatment of the individual 
who has a poor accident record. That is, the tests are used as a 
method of diagnosing the reasons for high accident liability in acci- 
dent repeaters. It is a process of individual treatment and recommen- 
dations based upon the particular individual's test scores. 

DeSilva has reported the results of this kind of clinical diagnosis 
and advice in a group of accident repeaters who were followed for 
a period of nine months after they had passed through the ''driver 



[Ch. 12 

clinic." Out of one group of 180 repeaters who came to the cHnic 
in one city, 101 were tested and given advice, while the remaining 79 
received some educational material of a safety campaign without the 
tests. Table 37 shows the accident records of these two groups be- 
fore and after the time of coming to the clinic. ^° An even more 
marked effect is reported for a group of 470 repeaters. The subse- 
quent accidents for a period of six months were 90 per cent fewer 
in the tested group than they were in a similar group which was given 
''no tests and no education." 

TABLE 37 * 
Re-educational Effect of Driver Test Clinics 

Comparison of accident record before and after clinic operation. Tested and 
untested repeaters. Nine months follow-up, Manchester, N. H., January, 1937 

Number of Accidents Accruing to Groups of Repeaters 
Accidents per 100 Drivers 

(N = 79) 

Before summons to clinic (15 months) 195 

After summons to clinic (9 months) 38 

Accidents after as per cent of accidents before 
9 months in each case 32.5 per cent 

Advantage of re-education by driver test clinic 
(compared to police attention, safety cam- 
paigns, etc., alone) 

♦H. R. DeSilva, The "why" of accidents, Traffic Safety, 1938 (July). 

Johnson and Cobb have questioned the value of DeSilva's evidence 
as a demonstration that drivers' clinics are effective in reducing the 
accident rate in treated drivers. Their criticism may be summarized 
in its application to the data of Table 37. The greater improvement 
in the tested drivers might be interpreted as ( 1 ) meaning that the two 
groups were different to begin with, and therefore might be expected 
to show different subsequent records; (2) that the clinical use of the 
tests is actually effective; or (3) that some other factor in the situa- 

30 No statement of the significance of this difference is given by DeSilya, and 
the present writer has been unable to discover a valid procedure for testing the 
significance because of the non-normality of accident distributions. Fletcher, whose 
similar studies are summarized on a subsequent page, reports a "critical ratio" for 
differences of accident rate, apparently basing his analysis upon standard statistical 
procedures which assume normality of the distribution of means. 

31 DeSilva and Forbes, Driver Testing Results, p. 49. 

3 2 H. M. Johnson and P. W. Cobb, The educational value of "drivers' clinics," 
Psychological Bulletin, 1938, 35, 758-766. H. R. DeSilva, Automobile drivers can 
be improved, ibid., 1939, 36, 284-285. H. M. Johnson, Evidence for educational 
value in drivers' "clinics," ibid., 1939, 36, 674-675. 

(N = 101) 


13.7 per cent 
57.5 per cent 


tion at the time of testing tended to reduce the accident rate.^^ With 
regard to the first of these possibiHties, Johnson does not consider 
that DeSilva has shown the equivalence of the two groups before the 
time of testing. Whether this criticism refers to the difference in 
average accident rate between the two groups, or to other possible 
differences in variability within the groups, he does not state. Even 
if the differences were accepted as statistically meaningful, however, 
Johnson would still be in doubt as to whether (2) or (3) is the cor- 
rect interpretation. 

It does not seem that the differences in mean accident rate are 
sufficient to account for the different subsequent histories. According 
to DeSilva, there was no special selection of the drivers to be tested, 
and all had been sent to the clinic in the same manner. There is still 
the possibility that the change was not produced by the testing and 
advice which was given. Perhaps any tests, valid or invalid, would 
have the same result, somehow focussing the attention of the driver 
upon his driving habits. This is a distinct possibility, but again it 
does not seem that it would be sufficient to account for the whole 

Fletcher has given more extended data upon the effect of the 
driver testing program.^* The study includes several groups of truck 
operators, groups of state employees, accident repeaters whose acci- 
dent rate had been increasing for a period of eight years prior to the 
test, and a number of drivers who had been warned because of re- 
peated traffic violations such as speeding. In each case a similar 
group which was not tested was tabulated for comparison. If John- 
son's criticisms of DeSilva are valid, they would probably apply also 
to Fletcher's studies. The comparison groups were not always strictly 
comparable in accident rate, and the distribution of accidents in each 
group is not given. In' addition, there is not enough information 
upon his methods of determining the significance of changes to permit 
us to evaluate these procedures. Nevertheless, the changes resulting 
from the testing are so marked and consistent from group to group, 
that it scarcely seems possible to regard them as artifacts rather than 
bona fide effects of the testing program. In the case of the accident 
repeaters, for example, the accident rate declines within a year of 

33 Johnson and Cobb assume that it is possible to test for significance of differ- 
ence of accident rates, but they fail to mention the fact that special methods are 
required because of the non-normality of the distributions involved. So far as the 
present writer has been able to discover, this statistical problem has not yet been 
adequately solved. 

3* E. D. Fletcher, Preliminary report on special tests (op. cit.). Part II. 
Effect of special tests on driving ability. (In mimeographed form.) 





the tests to a level lower than that for any of the preceding eight 
years (see Figure 35). 








1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 




100 c 










Figures 35. Accident Reduction Following the Use of Special Tests (Accident 
record of 144 accident repeaters) 

(From Fletcher, op. cit., Part II, p. 12.) 

The Generality of Accident-Proneness. — The topic of generality 
of accident-proneness has essentially the same meaning as the question 
of generality of any personal trait. It concerns the question of 
whether a person who is accident-prone in one occupation is also likely 
to be accident-prone in some other activity. Does the accident-prone 
driver also tend to have high accident liability in factory work ? Does 
the person who is prone to minor mishaps such as cuts and scratches 
also tend toward greater liability to serious or fatal accidents ? 

The importance of answering such questions should be clear. If 
accident-proneness is a consistent trait i-n many circumstances, indus- 
trial employment departments will need to take each applicant's total 
accident history into account in selecting workers for hazardous oc- 
cupations or even for occupations where the hazards are relatively 

Ch. 12] 



low. If proneness to different kinds of accident in the same work 
tends to be consistent in the same individual, safety engineers will 
wish to transfer workers who show trends toward minor injuries in 
order to avert more serious disaster. These are two distinct ques- 
tions concerning the generality of the trait. They will be considered 
separately in examining the available data. 

To give general answers to these questions would require a vast 
amount of research upon very large numbers of individuals over long 
periods of time. Consequently we can find only a few hints of the 
possible answers in the literature of the subject at the present time. 
We should have to raise these questions even if we had no informa- 
tion at all, however, since we must stress their importance as funda- 
mental to the understanding of the whole topic of accident-proneness. 
Until more information of this kind is forthcoming, our handling of 
the question must be full of gaps. 

Let us consider first the question of correlations between major 
and minor accidents within the same occupation or activity. Farmer 
and Chambers report correlations of this type based upon over four- 
teen thousand dock workers, including laborers, shipwrights, boiler- 
makers, and the like, with the data covering a period of one year.^^ 
In various groups the correlation between minor accidents and major 
accidents (causing absence from work for a day or more) ranged 
from zero to .36. While the numbers involved are large enough to 
make these correlations significant in some groups, their small size 
may be due to the shortness of the period of observation. Here again 
we are using accident rate as an index of accident liability. 
The major accidents, being more rare, must be studied over longer 
periods of time before accident liability and rate are closely 

Another more recent study also indicates a relationship between 
tendencies toward major and minor injuries. Jurgensen surveyed 
the data for 946 employees of one mill.^^ Among those who had 
disabling accidents, there was an average of 6.5 minor injuries in the 
year. Those who did not have disabling accidents averaged only 3.1 
minor injuries. A similar relationship was evident when a serious 
accident was defined as an accident requiring the care of a physician. 
Among those who had such serious accidents, there was an average 

35 E. Farmer and E. G. Chambers, A psychological study of individual differ- 
ences in accident rates, Industrial Health Research Board (Great Britain), 1926, 
Report No. 38. 

3® C. E. Jurgensen, How much do we know about accident causes? National 
Safety News, 1945, 52, No. 5, 26-27, 80, 82, 84. Abstracted in Psychological 
Abstracts, 1946, 20, 177, Abstract No. 1642. Original not seen. 



[Ch. 12 

of 5.5 minor injuries per person. Among those free of serious ac- 
cident, there was an average of 3.0 minor injuries. 

Until more complete data are available, it would appear to be 
best to assume that the two kinds of accident liability are closely cor- 
related until we have clear proof to the contrary. While these 
statistical data leave the possibility of close relationships open, there 
is an important logical reason for expecting a correlation.^^ The 
same unsafe act may at one time lead to a minor scratch, at another 
the circumstances turn it into a serious injury. An individual is not 
liable to accidents as such, but has tendencies or habits which lead 
to the performance of certain unsafe acts. In some occupations, 
therefore, there may be a very close relationship between serious and 
minor mishaps because the same unsafe act can result in either.^^ In 
other occupations, however, the actions which lead to serious injury 
may be quite different from those which lead to cuts and scratches. 
While it is likely that the correlation will vary markedly from one 
occupation to another, it is safest to assume a close correlation rather 
than ignore the problem until the correlation is demonstrated. It 
should do no harm to institute transfers, retraining, or whatever else 
is necessary to prevent the minor injuries, and we may be preventing 
major accidents by doing so. 

Another aspect of the interrelations of accidents concerns the cor- 
relations between accidents of different types in the same kind of work. 
For example, is there any tendency for a person who has a large num- 
ber of falls to be specially liable to accidents involving machinery, or 
to cuts and bruises in the use of hand tools? In their study of bus 
operators, already referred to (page 275), Farmer and Chambers 
determined the intercorrelations between five types of driving acci- 

37 Tiffin (op. cit., page 281) presents data to show that major and minor acci- 
dents are uncorrelated when the rates are determined by departments or occupa- 
tions. This is, of course, a distinct problem. The factors in a given job which 
lead to minor accidents are not the same as those which lead to major injuries, as 
Tiffin properly concludes. These results have nothing to do with the individual, 
however, but with occupational differences in hazards. 

38 H. W. Heinrich {Industrial Accident Prevention, New York, McGraw-Hill 
Book Co., Inc., 1941, pp. 26-35) has urged that an accident be defined, not in terms 
of injury, but in terms of events which might produce injury. From various 
observations he has estimated that for each major injury resulting from such an 
event there are, in the long run, twenty-nine minor injuries and three hundred 
occurrences which do not result in injury at all. While these ratios will probably 
,vary from occupation to occupation, the point is an important one. If records are 
kept in such manner as to include all falls, slips, and the like, their cause may be 
noticed and eliminated before the injuries occur. Heinrich was referring to locating 
causes of accident which would involve faulty tools, layout, lighting, or any re- 
sponsible factor. The same reasoning may, however, be applied to the individual 
factors in accidents in many instances, 

Ch. 12] 



dents. The results of this analysis are given in Table 38. These 
authors conclude that some of these correlations are significant, and 
that there is a slight tendency for the different kinds of accident to 
be related to one another. Since the groups of drivers were not large 
to begin v^ith, and the time period for Group B v^^as only two years, 
it is quite possible that the small size of these correlations is due to 
random variations which would be eliminated with larger numbers 
and longer time-spans. 

TABLE 38* 

Product-Moment Correlation Coefficients Between Different 
Types of Accident 

Group A 

Group B 


- 0.021 



Errors of judgment and miscellaneous 


- 0.001 



Overruns and skids 











Miscellaneous and blameless 



♦Farmer and Chambers, op. cit., 1939, Report No. 84, p. 12. Reproduced by permis- 
sion of the Controller of His Britannic Majesty's Stationery Office. 

Herdan has suggested the application of factorial analysis to the 
interrelations of accident records.^® For those familiar with Spear- 
man's methods of factorial analysis, we may mention Herdan's results 
briefly. He first points out the similarity between determining the 
liability of an individual to accident, and determining the *'G," or 
general factor, in intelligence tests. When the Spearman method 

3 9 G. Herdan, The logical and analytical relationship between the theory of 
accidents and factor analysis, Journal of the Royal Statistical Society, 1943, 106, 

^0 In fact, all the problems which we are discussing in this section could be 
treated by a variety of factorial methods, were sufficient data available. Spearman's 
methods would be satisfactory when they are applied to the problem of determin- 
ing gross accident liabilities, with no need to difTerentiate between different kinds 
of accident. Where we are interested in determining whether or not there is a 
general factor underlying accidents of a number of different kinds, or in a number 
of different situations (which is the basic problem we are concerned with here), 
there is no one technique which can be said to be "correct." Thurstone's methods 
could be applied just as well, and group-factors rather than a single "G" factor 
might appear. Nevertheless it would appear that the field of accident statistics 
could profit by extended factorial studies using various techniques. 



[Ch. 12 

is applied, he finds that the miscellaneous accidents are most heavily 
''saturated" with the general factor of accident liability. Next, and 
equal, come the errors of judgment and the accidents in which there 
was no blame to the individual. Herdan considers this reasonable, 
since an error of judgment may complicate matters in a situation 
where the other driver is technically at fault. In other words, an 
individual may show his tendencies to accident in some accidents 
where the other driver is still at fault. The accidents involving skid- 
ding showed the lowest relationship to the general liability factor. 

Herdan's results can, however, be considered as no more than an 
illustration of the possibilities of a method, since the data with which 
he works are unreliable, and the classification of accidents is probably 
not very satisfactory. In addition, other techniques of factorial 
analysis might have resulted in an entirely different picture of the 

The relation of accident liabilities for the same individual in dif- 
ferent occupations is still less known. One study has shown a cor- 
relation between the driving accidents of truck operators and their 
personal-injury accidents. " 'Personal accidents' included not only 
those which occurred as the result of collisions in traffic, although 
these constituted an appreciable fraction of the total number. The 
larger proportion, however, were accidents which occurred in daily 
work not directly connected with driving ; for example, many of these 
drivers were repairmen. Many of their accidents were caused by 
falling off their trucks ; by dropping tools on their toes or other parts 
of the body; by mashing, burning, or cutting fingers; in general, 
accidents due to personal clumsiness." Table 39 shows the nature of 
this correlation of two types of accident very clearly. 

Once more we find that our evidence is very incomplete. In this 
problem also there are logical reasons for expecting some degree of 
correlation between accidents of the same individual in different oc- 
cupations. Certain defects of vision would have their effects in a 
number of different situations. The same might be said of almost 
any deficiency which is related to accident liability in any one field. 
A few deficiencies, however, might be specific to a single activity. 
Night blindness, for example, could not be expected to bear a rela- 
tion to accident liability in most factories. Thus it is likely that a 
high degree of correlation of accident tendencies will be found in 
studying some pairs of occupations, while other pairs will show little 

41 U. S. Bureau of Public Roads, The Accident-Prone Driver, op. cit., pp. 31, 32. 


TABLE 39 * 

Personal Accidents of Drivers for a Public Utility Company 
Compared with Motor- Vehicle Accidents of the 
Same Drivers: 1930-1935 

(Average personal accidents per operator grouped in accordance with the 
number of motor-vehicle accidents each in the same period.) 

Number of Motor-vehicle 


Personal Acci- 


Accidents per Operator 

of Opera- 

dents per 























5 and more 






* U. S. Bureau of Public Roads, op. cit., p. 32. 

Again, however, it would be best to assume that there is correla- 
tion until it has been proved otherwise. An individual with a record 
of a high rate of accidents in any field should not be placed in a 
hazardous occupation until more information is available. The fur- 
ther information which we would need would be either the demon- 
stration that there is no correlation of occupational accidents in the 
fields involved in this particular case, or test results which show that 
the individual's deficiencies will not prove a handicap in the occupa- 
tion for which he is being considered. 

"Personality" and Accident-Proneness — It is frequently sug- 
gested that certain emotional or temperamental traits contribute to 
accident liability in certain individuals. In fact it is sometimes im- 
plied that most accident-prone individuals are high in liability because 
of these general and ill-defined "personality" traits, rather than be- 
cause of specific deficiencies in sensory or motor equipment. If this 
is true, we certainly should expect a high degree of generality of the 
trait of accident-proneness, since it would depend very little upon the 
specific activities and performances of any given job. It would also 
mean that we have a long way to go in developing means of diag- 
nosing accident-proneness. 

We are not in position to say how much of a part these "person- 
ality" characteristics play in accident liability, since we do not have 
as yet any clear statement of what personality traits are involved, 



[Ch. 12 

nor do we have any very satisfactory personality tests to work with. 

There are suggestions of the importance of temperamental factors 
in some of the available accident statistics. Vernon found, for ex- 
ample, that accident rates of individuals correlated with their visits 
to the first-aid room for treatment of minor ailments, although there 
was no correlation between rate of serious illness and accidents. 
Since minor ailments are frequently correlated with neurotic tend- 
encies, the latter may be the responsible factor in determining the 
correlation. Still more indefinite clues appear in the clinical studies 
of the accident-prone in which there is always a group which is classi- 
fied under the heading "faulty attitude," the investigators having 
failed to discover any more definite causes of accident tendency. 

In recent years Dunbar and her collaborators in the field of psy- 
chosomatic medicine have been collecting psychiatric case studies of 
hospital patients under treatment for fractures.*^ Originally these 
patients have been chosen as a "normal" control group for studies of 
other groups of patients, e.g., those suffering from heart disorders. 
The investigators found, however, that many of these patients were 
accident-prone, and that they displayed a high frequency of previous 
accidents. They also came to the conclusion that there was a com- 
posite personality pattern which could be derived from analysis of 
their case studies of fracture patients, a pattern distinct from that 
found in other groups of patients. 

Before summarizing the details of personality characteristics of 
the accident-prone group, it would be well to examine the point of 
view of the investigators as they approach their data. Their conclu- 
sions must be based upon voluminous notes of interviews and 
personal histories. There is, therefore, a definite chance of the in- 
vestigator's becoming set to look for certain elements and thereby 
oversimplifying or distorting the results. 

These investigators assume first of all that there is proof that 
tests based upon co-ordination, reaction time, driving skill, and the 
like are invalid. The data which we have presented would justify no 
such assumption. It is true that the validities so far observed have 
been low — in some studies much lower than in others. Nevertheless 
we have already given reasons why these validities would be low 
even if all accident-proneness were due to specific deficiencies like 

4 2 H. F. Dunbar, T. P. Wolfe, E. S. Tauber, and A. L. Brush, The psychic 
component of the disease process (including convalescence) in cardiac, diabetic, 
and fracture patients : Part II, American Journal of Psychiatry, 1939, 95, 1319- 
1342. F. Dunbar, Medical aspects of accidents and mistakes in the industrial army 
and in the armed forces, War Medicine, 1943, 4, 161-175. H. F. Dunbar, Psychoso- 
matic Diagnosis, New York, Harper & Bros., 1943. 

Ch. 12] 



poor dexterity or reaction time. Poor validity of these tests does not 
prove that the factors causing accident-proneness must be sought in 
general personality characteristics. Dunbar appears to have exagger- 
ated the poorness of the relationships in previous studies by concen- 
trating her attention upon the negative findings. This assumption 
does not necessarily lead to outright errors, since all would admit 
that general personality factors will be important in a certain pro- 
portion of accident cases. It may lead, however, to a mistaken em- 
phasis upon these temperamental characteristics at the expense of 
careful study of other possible factors. 

We should have to know more details of the methods used by 
Dunbar and her co-workers if we were to weigh the possibility that 
a ''type" has been imposed upon the data by selection. Although 
Dunbar speaks of certain subtypes of accident-prone individuals 
(based upon differences in diagnosis by the Rorschach Test), the 
picture which is presented is that of certain common features run- 
ning through most of the group, with the subtypes representing 
variations in certain specific traits. We should have to see more 
definite quantification and statistical treatment of the distributions 
of these traits before we could determine the consistency of the pat- 
terns. As a simple example, suppose it is found that the fracture 
patients show two traits in unusual frequency; very strict parents 
and tendency to truancy from school in childhood. Does this mean 
that there is one group in which the first trait is very frequent and 
is related to accident frequency, while there is another distinct group 
of individuals who display the second trait in high frequency? Or 
does the result mean that the fracture patients tend to show a pattern 
in which the two traits appear in combination in abnormally high 
numbers? The investigators applied statistical treatment of indi- 
vidual traits, but the discovery of patterns of traits appears to have 
been carried out on the basis of general impression, coupled with 
certain theories of personality organization derived from psychiatric 
practice. Yet the patterning is as important as the discovery of single 
traits, since entirely different methods of diagnosis and prediction 
would have to be applied in the two cases. 

With these cautions in mind, we may attempt a brief summary of 
the personality traits which Dunbar found to be correlated with being 
a member of the fracture group. Since most of the comparisons 
involve the prevalence of the trait in the fracture group as compared 
with the group suffering from coronary occlusion, some of the 
''traits" may not be special characteristics of the fracture group. In- 
stead, they may frequently represent traits in which the coronary 



[Ch. 12 

group is abnormal, while the accident group is normal. In some 
instances it is possible to refer to a general norm; in others, no 
general norm is available. 

The following are some of the more clear-cut distinctions between 
the two groups, briefly noted : 

1. Higher accident history in family and siblings of accident cases. 

2. Better health record. 

3. Tendency to interrupt education, although intellectual 'level is 

4. Some accident-prone tend to stick to jobs, but without attempting 
to advance; another group shows job mobility. Coronary group 
sticks to one job and tends to "work to the top." 

5. Higher proportion of early neurotic traits, such as lying, steal- 
ing, truancy, sleepwalking. Later, almost no obvious neurotic 
traits. Reverse tendency in coronary patients. 

6. Tendency to take stimulants to "let off steam," while coronary 
group may take them to keep on working. 

7. Greater interest in sports among fracture patients. 

Beyond these relatively specific comparisons, Dunbar also presents 
a more qualitative differentiation of temperamental characteristics. 
She depicts the accident-prone individual as being impulsive — the 
tendency to do something when involved in an emotional situation. 
The cardiac patients did not show this tendency. The accident pa- 
tients try to escape from authority, including that of strict parents, 
while the cardiac individuals try to overcome or subdue the authority. 

Dunbar has proposed the use of selective procedures in the armed 
forces based upon the patterns we have just sketched. The informa- 
tion would be gained from the history of the individual, from ques- 
tioning him, and from observing his behavior during induction or 
training.*^ Thus far, however, the predictive value of such a pro- 
cedure is still to be determined. It is one thing to find distinguishable 
patterns of traits among those already known to be accident-prone, 
it is quite another thing to pick out the accident-prone from a heter- 
ogeneous group of inductees or job applicants. 

The work of Dunbar is the only study of this problem of a sys- 
tematic kind.** In the literature on psychoneurosis there are 

^3 op. cit., War Medicine. 

Johnson {op. cit., Psychological Bulletin, 1946, 518-524) refers to the psychiat- 
ric approach to accident problems, pointing out the lack of even elementary statis- 
tical analyses in instances of this approach which he has observed. He does not, 
however, take account of Dunbar's work in his discussion. 

Ch. 12] 



occasional references to individuals who had accidents because of 
some particular neurotic symptoms. Culpin describes several such 
cases in a discussion of accident-proneness, but the cases came to his 
notice incidentally in general psychiatric practice.*^ Hersey also gives 
instances of "emotional" causes of accidents and accident-prone- 
ness.*® Dunbar's analysis would imply that the ordinary kinds of 
recognized neuroses are not an important factor in the adult accident 
repeater. If this is correct, devices for diagnosing neurotic tendency 
would not be effective in weeding out the accident-prone from dan- 
gerous occupations. Nevertheless it would be worth while to try out 
these diagnostic methods when and if measures of satisfactory re- 
liability and validity are available. Certain kinds of neurotic pattern 
may entail accident tendencies as a part of the expression of the 
neurosis. Others may be accident-prone for temperamental reasons, 
but not classifiable as neurotics in the customary meaning of the 
term, like the cases described by Dunbar. 


Needs for Research in Accident-Proneness. — As in most fields of 
research, when our knowledge of a field increases, we become more 
and more aware of the important gaps in our information, and of the 
need for new types of research which could not be envisaged before 
our present knowledge was obtained. Some of the information 
which we need can be gained from further statistical analysis of ac- 
cident data itself. Other problems require the development of more 
adequate procedures of analysis of individual differences, and fur- 
ther study of their correlations with accidents. 

The information which we need to derive from the accident 
statistics consists largely of correlations. As we have seen, we wish 
to know more of the generality of the accident trait. Does the indi- 
vidual tend to be prone to only one kind of accident, or is he prone 
to accidents in general? Does he tend to have accidents only in a 
certain kind of situation, or do they occur in a great variety of 
circumstances? Is the seriousness of the outcome of an accident a 
matter of chance, or does accident-proneness express itself in leanings 
toward minor accidents in abnormally high or low proportion? 
That is, does accident liability express itself in tendency toward a 

45 See the discussion of Newbold's paper in the Journal of the Royal Statistical 
Society, 1927, 90, 540-544. 

*6 R. B. Hersey, Emotional factors in accidents, Personnel Journal, 1936, 15, 



[Ch. 12 

certain frequency of accident without regard to the seriousness of 
the result, or can a person have a high HabiHty to minor accident 
coupled with a low liability to serious accident ? As we have seen, we 
already have some hints on these problems, but there is a field for 
much profitable research in extending our knowledge of these 

With regard to the individual characteristics which may be cor- 
related with accident tendencies, we have already seen that the pro- 
cedures so far in use have been relatively crude — correlating test 
scores against over-all accident frequency. Some of the difficulties 
we have found in these procedures might be avoided by more elabo- 
rate methods of analysis. For example, Fletcher's report upon an 
analysis of intersection accidents, mentioned on page 276, suggests 
a pattern of analysis which would make test data much more meaning- 
ful. Instead of comparing accident repeaters en masse with accident- 
free groups, it would appear to be much more profitable to subdivide 
the accident repeaters into various subgroups and work out the re- 
lationships separately for each group. For example, the first step 
might be a study of those involved more than once in a single type 
of accident (intersection, collision in passing, running off the 
road, and such).*^ These divisions should be made as specific as 
possible. Then, if several types of accident appear to be correlated 
with the same pattern of test scores, these accident types could be 
grouped together. 

Another approach would consider the problem of patterning of 
deficiencies as a factor in accidents. It is conceivable, for example, 
that a person who is somewhat deficient in two traits, such as reaction 
time and speed estimation, is much worse as a driver than a person 

*7 It may be suggested that Cobb has already tried this kind of analysis without 
success. (See page 280 above.) We have seen, however, that his analysis was 
incomplete. He tested only for correlations which appeared "reasonable" on a 
common-sense basis. There may be other relationships which could not be guessed 
in this manner. The proposal made here is that the results of many different tests 
be examined for each type of accident. Certain individual tests or certain com- 
binations of tests might have a predictive value which is greater for the specific 
kind of accident than it is for accidents in general. 

Another difference between this proposal and Cobb's method is that Cobb cor- 
related scores on a given test with frequencies of accidents of a certain kind. We 
suggest here a selection of individuals who seem to be consistent in the kind of 
accident which they suffer, and a comparison of their test scores with those of an 
unselected group. Just as accident rate is partly a function of chance, so the type 
of accident may not always reflect the personal tendencies of the individuals involved. 
If an individual has two or more accidents of a similar kind, however, a personal 
factor determining the kind of accident is more likely. Such an analysis would be, 
in effect, a search for specific factors in accident tendencies, as contrasted with 
the problem of more generalized accident-proneness. 

Ch. 12] 



who has a single severe deficiency. Are there any patterns of de- 
ficiency which are related to very high accident liabilities? The 
analyses now available do not answer such a question. In the Har- 
vard clinics, for example, each test was analyzed separately. Farmer 
and Chambers used combinations of tests in their predictions of ac- 
cident rate (see page 275) but this was merely an additive combina- 
tion. There is no way of determining from such an addition of 
scores whether a particular combination of deficiencies has an effect 
which is more than the sum of the deficiencies. 

Dunbar's suggestions about personality factors will need to 
be followed up with further validation and more concrete pro- 
cedures for diagnosis if we are to take the "personality" factors into 
account in predicting accident tendencies. As she herself has sug- 
gested, there is no adequate control group of low-liability individuals 
which can be compared with the accident-prone with respect to the 
psychiatric patterns she describes. The ability of a psychiatrist to 
pick out of an unselected group those who have an accident tendency 
has not as yet been demonstrated. The therapeutic value of treating 
these cases has not been demonstrated either; in fact, there is less 
evidence for improvements from this form of treatment than there is 
for improvement as the result of specific aptitude testing. 

This brings us to a final need. This is the more thorough testing, 
in a statistically unimpeachable manner, of the therapeutic value of 
driver tests, and the trying out of similar clinical techniques for ac- 
cident repeaters in industry. 

All these suggestions call for years of research by a variety of 
specialists. It is hoped that the interest of governmental agencies 
and of industry in these problems will continue and develop, and that 
those who are working in the field will not overlook new approaches 
in the form of new testing methods which have not heretofore been 

Suggestions for Practice. — The knowledge of the problem of 
accident-proneness which we now have can contribute to practical 
approaches to accident prevention, even if there are many loopholes 
in our information. The safety department of a plant, the personnel 
department concerned with transfers of workers or hiring new em- 
ployees, the state highway authorities, can not only support research 
in the future but they can also make use of what has been done. 

H. F. Dunbar, Psychosomatic Diagnosis, New York, Harper & Bros., 1943, 
p. 690. 



[Ch. 12 

The following procedures are recommended as in keeping with 
our present knowledge of this field : 

1. Record keeping for accident reduction. 

(a) Detecting the accident-prone. As we have suggested, it is 
safest to assume generality of the accident trait. If we assume 
that a person with a high rate of minor injury is more likely 
to have serious accidents, such a person should either be trans- 
ferred to a less dangerous position or given remedial treat- 
ment if that is available. The personnel department should 
make every attempt to find out the accident history, major 
and minor, of a prospective employee, including off -the- job 
as well as industrial accidents. 

(b) Detecting external factors in accidents. Accident records 
should be kept not only in terms of the individual involved, as 
we have just suggested, but also in terms of the location, the 
cause, and all other pertinent surrounding conditions. As 
Heinrich suggests, this may also prevent serious accidents 
later by showing up factors in the working situation which 
make for "unsafe acts.'* 

2. Use of all available testing methods for hiring men to work in 
occupations with a high injury rate, and for diagnosing those who are 
already employed and who show an accident tendency. 

This would provide further information for validating the tests, 
and in addition would have a beneficial effect upon the worker taking 
the test. The tests might have this latter effect even if they have low 
validity — the worker may give more attention to possible habits or 
shortcomings which are making him accident-prone. 

Chapter 13 


In a general survey of the topic of efficiency, learning is both a 
factor contributing to efficiency of performance and an activity which 
itself provides problems of economy in time and effort. There is 
little doubt that practice results in marked improvements in efficiency 
of any given performance, provided that the worker is being taught 
effective methods of doing the job. In muscular work the metabolic 
cost per unit of output is lower in the skilled athlete or laborer than 
it is in the novice. Where indices of effort in sedentary work have 
been available, similar results have been obtained. In fact it is 
scarcely conceivable that practice could reduce efficiency unless the 
training is so ill-conceived that the worker is learning methods which 
are awkward and wasteful. Even when the method of work which 
is being taught is not the best method which could be developed, the 
practiced worker carries through the procedure with fewer waste mo- 
tions, fewer interruptions and errors, and he feels less strained. 
Again, accidents are usually very frequent during the novice stage 
of practice. 

Practice and training themselves require effort and take up the 
time of the worker and his instructor. Since improvement in per- 
formance involves a cost to the trainee, we are again faced with 
problems of efficiency. We wish to determine, therefore, the method 
of practice and the method of instruction which produce the maxi- 
mum improvement for a given amount of effort and time. 

In most writing upon the subject of efficiency of learning, the only 
element of cost which is taken into account is the time consumed in 
learning. In dealing with work in general we have found it a serious 
misuse of the term to confuse efficiency with rate of performance. 
While it is still misleading to confuse rate and efficiency of learning, 
there is probably more justification for the interchange of terms in 
connection with learning than there is in other aspects of work. For 
one thing, the trainee is using not only his own time and effort, but 
usually the time and effort of an instructor also. If the trainee 
adopts a method of learning which is slow but easy for him, it may 




[Ch. 13 

require more effort by the instructor. In any case, the time factor 
becomes doubly important. Secondly, the main danger in overlook- 
ing the total cost of work in the ordinary working situation is that 
the additional cost may lead to an accumulated deficit, as also to a 
long-term deterioration of the worker, with ultimate loss of produc- 
tivity. During the training period, however, it may require more 
effort for the trainee to learn quickly, but this is only a temporary 
condition which is not likely to lead to harm in most jobs. Then, too, 
the trainee arrives at a stage of practice where the task itself, as dis- 
tinguished from the practice, requires less and 1-ess effort. In other 
words, inefficiency due to high effort level in early training is a 
temporary inefficiency which may be more than balanced by the later 
efficiency of the worker. 

For the above reasons we do not need to be seriously concerned 
about the confusion of rate and efficiency of learning. The confu- 
sion does exist, however, and it would be clearer and more honest 
to speak of the rate of learning in discussing most of the research 
that is available in this field. We certainly know very little about 
the true efficiency of learning or the factors which affect it. 

The amount of published material upon the subject of methods 
of practice and training is enormous. It consists of information 
falling into several categories: (1) Laboratory studies of single 
factors in isolation. (2) Factory experiments in which single vari- 
ables are studied but others are only partially controlled. (3) Ex- 
periments in the schools dealing with the learning of academic 
subjects. (4) Discussions of ''practical" training programs in which 
the principal emphasis is upon the administration of the program, 
e.g., such topics as how the instructors are chosen, the content of 
training courses, and the like. 

Although each kind of information has its special value, the 
various kinds are by no means equal in usefulness. The fourth group 
of ''practical" discussions appeals to industrial managers and super- 
visors, but frequently it tells us almost nothing about how the rate 
of learning can be improved. Often there are no quantitative data 
to show the effect of the training program. When such data are 
given, the program is so complex that we cannot tell what factors 
were responsible for the improvements which occurred. Often the 
procedures which are supposed to have increased the speed of learning 
are special tricks which would apply only to that special job. They 
are therefore of importance only to those who are training workers 
for the identical occupation. 

Ch. 13] 



Of the other data that are available, the scholastic experiments 
frequently involve a situation in which the motivation is so different 
from that common in industrial training that it makes it extremely- 
difficult to transfer the results. The same criticism also may be ap- 
plied to a lesser extent to the laboratory experiments. In any case, 
these methods provide at most a suggestion which must be checked 
later by direct factory experimentation. 

The ideal approach to the development of an industrial training 
program would consist of drawing up recommendations from the 
material already available in the field of learning, and then testing 
those recommendations by means of an experimentally controlled 
trial. In more detail, the procedure might be outlined as follows : 

L Determination of the best method of performing the work; that 
is, finding out in detail what is to be learned. This problem was 
considered in Chapter 7. 

2. Determination of the sequence in which the aspects of the job 
are to be taught, ignoring the traditional method of teaching, and 
considering experimental information which might have a bear- 
ing upon the ease of learning the material in various orders and 
in various methods of subdividing the job. 

3. Drawing up a detailed teaching and practice plan. This plan 
would contain: 

(a) Descriptions of the instructor's technique in introducing 
the worker to each phase of the job. 

(b) Scheduling the practice sessions, including their dura- 
tion and spacing, and the type of practice which is to be 
carried on during each session if possible. 

4. Explicit provisions for motivating and developing interest. (E.g., 
instructional material which would make the job meaningful even 
though the material is not essential to the operation in itself, 
provision for incentive pay during training, and the like.) 

5. Selection of a suitable measure of proficiency. (If the perform- 
ance itself is not reliably measurable, it will be necessary to use 
other forms of achievement test. See a general textbook on in- 
dustrial psychology for a discussion of achievement testing.) 

6. Provision of a control group to be trained by former methods, or 
by other methods under consideration. The control group of 
workers should be as nearly identical with the experimental 
group as possible in each of the following qualifications: 

(a) age, 

(b) aptitude for the work, 



[Ch. 13 

we mean by recollection, the latter is simply a result of rote mem- 
orizing. Although recollection is not an explicit aim of most forms 
of industrial training, it is an aid which the learner frequently uses 
— he recreates the diagrams of his textbook, he listens again to the 
instructions of his teacher, and so on. It is a component of learning 
which we cannot now control because it is so little understood. It 
appears to follow completely independent rules, and to be dependent 
upon many factors other than those which control the four types of 
learning we have listed in Table 40. It must be kept in view because 
it may account for some of the apparent irregularities which occur in 
learning. Individual differences in learning under a given set of con- 
ditions are often due to the differences in individual ability to recol- 
lect in this manner. Also, two tasks which appear to be very similar in 
nature may not respond to the same conditions of learning because 
of the different involvement of recollection in the two situations. 

Kinds of Learning 

1. Verbal memorizing. 

2. Learning of a fixed motor pattern. 

(For many purposes this class may be grouped with the verbal type of 
memorizing. Examples would be typing and machine operation of some 

3. Learning dependent upon comprehension. 

(The task involves mechanical principles, understanding of an account- 
ing system, or the like.) 

4. Learning an adjustive or variable motor pattern. 

(Operating a car or airplane, learning a sport like tennis. What is 
learned is not a fixed sequence of movements but a pattern which 
constantly varies with a changing situation.) 

As we proceed to consider the things which the training super- 
visor must keep in mind as he plans, we shall discuss only those factors 
which have been investigated experimentally. We shall have to con- 
sider not only the presence of experimental evidence in support of a 
given recommendation, but also the extensiveness of this experimen- 
tation in the various classes of learning we have described. A factor 
which has been found to be beneficial in a number of experiments 
upon rote memorizing may legitimately be recommended for appli- 
cation to other tasks involving rote memorizing, subject to the ex- 
perimental checks which we have already described. To extend the 


recommendations to tasks involving other classes of learning is a 
more doubtful procedure. It is worth the trial in many cases because 
such a trial extends our knowledge of the subject of learning. In 
terms of the immediate practical aims, however, such a recommen- 
dation is not to be considered much more valuable than ordinary 
common-sense hunches and guesswork. 

On preceding pages we have outlined the task of the person who 
must plan a training procedure. In presenting the evidence upon 
factors in learning, we shall follow this same outline. It would be 
easier to survey what is known without reference to the details of the 
planner's task, and it would give the impression that a substantial 
amount of information is available. Even though there is a back- 
ground of hundreds of experiments upon learning, there are many 
gaps in our knowledge which appear clearly only when we try to 
match our experimental results with the task of the training super- 

Our approach may not do justice to the body of scientific 
information upon learning now available, but it will show the appli- 
cability more clearly, and it will also bring out the need for much fur- 
ther research. 

Factors in Determining the Subdivision and Sequence of Ma- 
terial. — It might appear that most subjects to be taught, whether in 
a factory or in a school, have natural subdivisions and natural orders 
in which the various subdivisions may be taught. It is true that 
many of these divisions and sequences are fixed by the nature of the 
task to be learned. Many so-called ''natural" divisions, however, 
may be only the result of habit and tradition on the part of the 
teacher. Under these conditions, the planner of a teaching program 
should not be misled by the apparent naturalness of a sequence into 
ignoring other non-traditional possibilities which may be more ef- 
fective than those now in use. For example, it might be that the 
traditional method of teaching a person to drive an automobile by 
spending a long time in practicing the manipulations of clutch and 
shift lever at the beginning of practice is not the most effective one. 
Perhaps the person would learn more rapidly if he were to spend very 
little time on these manipulations at first, and were to spend more 
time practicing steering and driving on the open road before per- 
fecting his skill in shifting. We do not know which would be the 
more effective until both have been adequately tried out. Belief or 
disbelief upon the part of practiced teachers in the effectiveness of a 
method is insufficient evidence to settle the question. 



[Ch. 13 

Many problems of this sort can be settled only by research upon 
the job in question, following the outline we have suggested. Certain 
general features of subdividing the learning material have, however, 
been investigated in a sufficient variety of circumstances to permit 
some generalization. These generalizations should provide at least 
some background for operation by the person charged with laying out 
a training program. 

Size of the Practice Unit. One general factor which has been 
investigated quite widely is usually labeled as the ''whole-part prob- 
lem." In almost any task of learning, there arises the question as to 
how much of the material should be mastered before the trainee passes 
on to the next segment. In learning a language, should the individual 
master thoroughly a very few words, then a few more, and so on, or 
should he be given a large number of words all at once? If a job 
requires co-ordinated manipulations with both hands, as in playing 
the piano, should the trainee master the technique with each hand 
separately before attempting the combined operation? 

This question has been investigated with verbal lists, poems or 
numbers to be memorized, and also for certain muscular skills. In 
most cases there is an advantage for the "whole" method, that is, for 
exposing the trainee to a large amount of information, or to a more 
complex co-ordination, rather than subdividing material into parts 
which are (individually) more easily mastered.^ There are limits, 
of course, to the amount which can be placed in a learning unit, but 
the optimal amount of material, as determined in experimental 
studies, has been larger than that which would naturally be chosen by 
the trainee or by most teachers. It should also be remembered that 
the "whole" method can often be combined with special practice upon 
"parts," with results which are superior to those obtained with the 
complete emphasis upon the whole task. 

Although the principle of "whole" practice is quite general in the 
sense that it has been found effective in a variety of learning tasks, 
there are certain conditions in which the "whole" method becomes 
less effective. In verbatim memorizing it appears that the "whole" 
method is itself a skill which must be practiced. Subjects in some 
experiments showed an advantage for "part" practice early in the ex- 
periment, but later the "whole" method gave better results. There is 

1 For a survey of experimental results see R. S. Woodworth, Experimental 
Psychology, New York, Henry Holt & Co., 1938 or J. A. McGeoch, The Psychol- 
ogy of Human Learning, New York, Longmans, Green & Co., Inc., 1942. 

Ch. 13] 



also some indication that children profit less by ''whole" practice. 
This point may be especially important in certain kinds of industrial 
training where the trainees are unaccustomed to systematic learning. 

Most of the experimentation upon the factor of ''whole" learning 
involved tasks which belong to the first two types of learning men- 
tioned previously (page 302), and some refers to the fourth kind. 
Little or no experimental data are available about the effect of such 
a factor in learning which is dependent upon understanding. It is 
often assumed that the "whole" method should be applied to this 
kind of learning as well; for example, in scholastic work. Such an 
assumption is not justified, since there is no reason to suppose that 
this kind of learning depends upon the same factors as other kinds. 
Probably we shall ultimately find that the "whole-part problem" is 
much more complex when we turn to comprehending. 

Nevertheless, we shall probably find that some modification of the 
"whole" method is a good general rule even in learning that is de- 
pendent upon understanding. Each segment of a field is more mean- 
ingful and more readily learned if the student has at least a general 
notion of the place of that segment in the total field, and some idea 
of the interrelations between the thing learned and the other elements 
in the same total framework. 

We must emphasize that the preceding statements are speculative, 
and this is not supposed to be a speculative book. The speculations 
are inserted, however, in order to suggest the need for more adequate 
research upon the conditions of learning which is dependent upon 
understanding, certainly a very important form of learning not only 
in the school but also in industry. 

Factors in the Sequence of Training. Variations in the 
possible sequence of learning different aspects or segments of a job 
are potential factors in the rate of learning, but they have been given 
only isolated consideration. In rote memorizing or in learning a 
fixed movement pattern, the sequence itself is the important thing 
to be learned, and usually cannot be varied for that reason. In adap- 
tive learning and in learning dependent upon understanding, and 
even in the more mechanical forms of learning where the task con- 
sists of several distinct phases, sequence is a potential variable. Un- 
fortunately, these are exactly the kinds of learning which have been 
most difficult to study experimentally. 

One problem which arises where there is a choice of sequences is, 
however, amenable to experimental study. This is the problem of 



[Ch. 13 

transfer and interference. If, for example, it is decided to train the 
worker in the simpler phases of the job first, we must consider 
whether this initial training is likely to be an advantage or a disad- 
vantage when the trainee comes to the later phases of his training. 
We may sometimes wish to begin the training with a simplified 
version of the job, on the assumption that this will make later practice 
in more complex operations much easier to master. Similar con- 
siderations arise when we consider the transfer of workers from one 
task to another. 

The positive transfer of skill from one manual task to another, or 
from one sensory discrimination to another, is much more limited 
than is believed by many laymen. The mere fact that one job re- 
quires dexterity in finger manipulation (for example) does not 
guarantee that the worker will learn a new task involving the same 
general type of bodily movements. 

Two major conditions for positive transfer have come out of ex- 
perimental studies. First, transfer occurs if a large proportion of 
the new task is made up of movement patterns which are identical 
with those in the task learned earlier. We shall presently give an 
illustration of the fact that mere general resemblance of the move- 
ments is not sufficient, and that identity must be interpreted in a 
very narrow or rigid sense. The second condition favoring transfer 
is the existence of common principles underlying the two tasks — if 
these principles are explicitly taught or discovered by the subjects. 
In other words, the trainee must understand that the principles apply 
to both tasks if they are to lead to transfer. 

An experiment by Cox illustrates these two points.^ The two 
tasks both involved assembling and disassembling parts of an electric 
fixture. The only difference was that the two tasks involved differ- 
ent parts of the fixture. Yet the experimental group which practiced 
first one task and then the second did not learn the second task any 
more rapidly than the control group which practiced only the second 
task. This was the result for bare ''practice" as contrasted with what 
Cox called "training." In the latter case, the experimental group not 
only practiced the first task but also was given explicit instruction in 
methods, such as recommendations upon the arrangement of ma- 
terials and the distribution of attention. With this additional factor 
of instruction in general principles of workmanship, a reliable trans- 
fer effect was obtained. 

2 J. W. Cox, Some experiments on formal training in the acquisition of skill, 
British Journal of Psychology, 1933, 24, 67-87. 


Thus we should be wary of attempts to prepare the trainee for 
complex manipulations by training him first in simpler manipulations, 
unless the simpler task is identical in movement pattern to portions of 
the more difficult job, or unless the simpler activity is used to teach 
explicitly important general principles involved in both. If the 
simpler task is introduced solely as a training device, and if it is not 
a part of the job as carried on in practice, we may seriously question 
whether the introductory training saves anything in the long run. 
In some instances the gain may be illusory, due to the apparent ease 
of beginning the complex task after this introduction, but the over- 
all time may be no greater if the trainee starts immediately upon the 
complex task without any introductory practice. The planner of a 
training program must seek to avoid such illusory gains and to search 
for the procedure which will give most adequate results in the long 

It may be argued on common-sense grounds that these illusory 
gains are important in building up the self-confidence of the worker 
and keeping him interested in the learning. Such a claim can be 
substantiated, however, only by direct trial. It can be just as logi- 
cally argued that initial difficulties in learning would have a bene- 
ficial effect upon motivation. Engineering students, for example, 
are proud of the fact that their work is difficult and that many cannot 
learn it. A method of learning which sacrifices speed of learning in 
order to make the progress of the trainee deceptively smooth might 
be less challenging and therefore less interesting than a method which 
introduces complexities early. One of the experimental problems 
awaiting the further study of learning is involved here, and the 
optimum condition is still unknown. 

In other phases of planning the sequence, the possibility of nega- 
tive as well as positive transfer must be kept in view. In general, 
the greatest negative transfer or interference occurs where the two 
things to be learned are closely similar in their general nature, yet 
contain some disparate elements. When tasks of this nature are 
practiced one after the other, the interference may be in either direc- 
tion, i.e., the second practice may disturb retention of the first skill, 
or the initial practice may slow down the learning of the second 

Such interference is also a function of the amount of time elapsing 
between the two tasks to be learned. In studies of interference with 
retention (retroactive inhibition), a short rest between the two kinds 
of practice is very effective in reducing the interference effect. Po- 
tentially interfering phases of learning in any training program 



[Ch. 13 

should therefore be separated by rests or by periods of completely 
distinct activity. Interference is also found to be greatest when the 
initial activity is only partially learned. We may therefore recom- 
mend that one of the two activities should be thoroughly practiced 
before the second activity is introduced to the trainee. 

Scheduling Practice Sessions. Experiments with verbal 
memorizing and with the learning of motor skills have shown that 
the length of practice sessions is an important determiner of the ef- 
fectiveness of practice. More is learned per unit of time spent in 
practice when a number of short periods are distributed over several 
days, rather than when all the practice time is massed together. The 
actual length of practice period which is optimal for a given task will, 
of course, depend upon the nature of that task. Nevertheless a rough 
generalization might be that the optimal length of period is usually 
shorter than that which would be chosen by a person who is unaware 
of the experimental results. Consequently there is a general tend- 
ency to err in the direction of too great length of session. 

Many times the optimal spacing of practice cannot be secured be- 
cause of practical demands. Distributed practice is more effective 
from the point of view of maximum use of practice time, but the 
practical instructor must also consider the adequate use of the re- 
maining time of the trainees. It might be that progress per hour of 
practice in a given job, for example, would be greatest if the trainees 
spent one-half hour twice a day in working at the task. It would, 
however, take them several months to learn. By practicing several 
hours a day they are able to learn the task within a few weeks. From 
the point of view of total time-span spent in training, the massed 
practice may therefore be more effective. 

Although some considerations like that given in the above ex- 
ample will cut across the recommendations for distribution of prac- 
tice, the experimental results are not to be ignored. There are a 
number of ways in which the general findings are important. If, for 
example, a task consists of several distinct components (for example, 
a job which requires operating a drill press, filing, measuring, layout 
of work), each component can be scheduled for short practice ses- 
sions distributed over days and weeks, since the remaining time of 
the trainee can be used in practice upon the other components. 
Spacing of practice does not imply that the trainee must be idle 
between practice sessions. 

There are even instances where it might pay to allow the trainee 
to work only half a day, even from the point of view of the total 

Ch. 13] 



time-span of the training period. For example, the writer once heard 
of a group of men learning radio code who practiced eight hours 
per day throughout the program. The men undergoing this training 
had already received their basic naval instruction, and were not 
scheduled for any other type of training at the time. Consequently 
it was assumed that they should spend full time in their instruction 
and practice upon this one skill. The optimum length of practice 
session for this task has not as yet been established, but what evidence 
we have indicates that eight hours is certainly too long for maximum 
effectiveness of practice time.^ In fact it is quite possible that this 
period is too long from any point of view. Perhaps the total gain 
per week of the program would have been just as great if the men 
had spent as little as three or four hours per day in practice. It is 
even conceivable that the total time-span of the program could have 
been shorter with reduced hours of practice per day. We do not 
know whether or not this is true, but the example illustrates the com- 
mon failure to make any examination of this question, and the as- 
sumption that improvement in skill is a uniform function of the time 
spent in practice. 

Motivational Factors in Learning. There is a consider- 
able body of evidence that repetition is not in itself a sufficient 
condition for learning. The evidence consists of experiments in 
which the subjects are engaged in an activity in such a way that they 
do not realize that the experimenter is studying their learning of that 
activity. Under these conditions, lack of any intention to learn the 
activity makes the repetition almost completely ineffective. Although 
there may be conditions other than the explicit intention to learn 
which make repetition effective, it is nevertheless clear from these 
experiments that the effectiveness of practice will vary more with 
motivational conditions than with any of the factors considered so 
far. Indeed, one possible explanation for the facts we have already 
considered, such as distribution of practice and the subdivision of 
material, may lie in the variations in interest which occur as a result 
of these factors in practice. 

Ahhough the importance of motivation cannot be overemphasized 
in any discussion of learning, nevertheless little can be added beyond 
what has already been said earlier in this book in our chapters on 

3 In a review of the literature upon code learning, Taylor mentions that two 
investigators have recommended periods of not more than an hour per day (upon 
the basis of tmcontrolled observations rather than experimental study), D. W. 
Taylor, Learning telegraphic code, Psychological Bulletin, 1943, 40, 461-487. 



[Ch. 13 

incentives and motives. We need only consider special phases of 
the problem which come up in the learning situation. 

It will be noted that the experiments in which practice was inef- 
fective did not involve lack of any motivation upon the part of the 
subjects. They were motivated to perform, but not to learn. In 
other words, the direction of motivation or intention is an important 
factor in the learning process. On the other hand, we do find in- 
stances in everyday experience in which we learn something without 
the intention to learn. We apparently learn it because we examine 
(perceive, comprehend, or inspect) it with interest, or we get some 
kind of enjoyment from the activity. In the experiments which re- 
sulted in little learning without intent to learn, the material itself had 
little meaning or interest to the subjects. 

This brief glimpse of the motivational factors in learning brings 
us back once more to problems of efficiency which have been dis- 
cussed before. We should like to know the relative cost of learning 
uninteresting material under extrinsic incentives such as pay, as com- 
pared with learning material which is intrinsically interesting. 
Common experience would lead us to expect that the former case 
requires both additional effort and time for learning. If experi- 
mental studies could show that these differences are large enough 
to be practically important, we should then have an answer to the 
question of how much the instructor should do to make the task in- 
trinsically interesting. 

The instructor in a particular training program is limited in the 
extent to which he can build up intrinsic interest in the work among 
his trainees. The ability to become interested is, after all, a function 
of many things which have happened to the trainees before they came 
to the present instructor. He can widen those limits by proper se- 
lection of men who are to enter the training program, but once the 
men are selected his use of the factor of interest is definitely re- 
stricted. Nevertheless he must make maximum use of this factor. 
Extrinsic incentives in the form of praise, reproof, or monetary 
bonuses can be only a partial substitute for this factor. They must 
be considered as such, and not given the central role to which they 
are sometimes assigned. 

Understanding in Learning. Understanding plays a multiple 
part in learning. In those motor and verbal tasks where the thing 
to be learned is a sequence or pattern — a series of words to be repro- 
duced verbatim, a series of movements to be reproduced in detail — 
understanding is not essential to the performance of the task. Even 

Ch. 13] 


here, however, the introduction of meaning obviously makes the task 
easier to learn. Other kinds of learning depend upon the compre- 
hension of the relationships and problems to be learned. 

Understanding may also provide its own incentive which leads to 
learning. The mere fact that something is comprehended by the in- 
dividual often seems to provide it with enough interest to ensure its 
retention. In other words, it may be that to understand something 
is in itself a factor in interest. In addition, of course, understanding 
can relate the activity to be learned to other fields of endeavor which 
have already acquired interest for the learner. 

To serve these functions, understanding does not have to be com- 
plete, technically detailed, or even rigorously correct in the sense in 
which it would be regarded by an expert in the given field. The in- 
structor must concentrate upon achieving a method of presentation 
which gives an understanding at a level which is satisfactory to the 
trainee, and which is sufficient to serve the functions listed above. 
Thus the radio technician trainee does not need a complete under- 
standing of the theory of electron emission. The conception of elec- 
tron emission which he is given may be subject to serious criticism 
by a professional physicist, but the trainee's conceptions may be quite 
adequate for his purposes. 

In these areas the skill of the instructor must consist in achieving 
a compromise between detailed and accurate technical knowledge and 
loose conceptions which are likely to lead to serious error. The diffi- 
culty which is met in many training programs is in finding or training 
instructors who know the technical basis of the job they are teaching, 
and yet are willing to forego complications in their instruction. 

Conclusions. — The recent war brought with it a great increase in 
the number of individuals specializing in the direction and planning 
of training programs. Though many of these specialists collected 
valuable experiences and observations, few of the observations were 
made under anything like controlled conditions. Many of those who 
directed training were themselves untrained for the work. Those 
who did have some background in educational method were too often 
attempting to extrapolate from what they had learned of schoolroom 
techniques to the problems of factory training. That would be satis- 
factory only if experimental studies were used to check upon these 
skilled guesses. 

This situation is apparently the reason why we have not seen very 
many reports of the controlled testing of training methods in industry 
in spite of the large amount of training which went on during the 



[Ch. 13 

war. Perhaps such studies have been carried out and have not yet 
been published for one reason or another. Yet the controlled study 
of training methods, with deliberate variation in the conditions of 
learning, with industrial tasks learned in the industrial setting, is 
the only sound way in which our techniques of training can be ad- 
vanced. Once more we come back to our much repeated point that 
industry cannot expect decisive aid from psychology until it actively 
maintains research programs directed toward the solution of its 
particular problems, not only its immediate, everyday problems, but 
also the broad fundamental questions which underlie them. 


Adams, S., 129 

American Society of Heating and Venti- 
lating Engineers, 136 
Arai, T., 67, 68 
Ash, 1. E., 60 

Atzler, E, 38, 39, 54, 61, 110, 164 

Baker, K., 131 

Balchin, N., 163 

Barkin, S., 222 

Barmack, J. E., 42, 202 

Barnes, R. M., 158, 159, 167, 213, 217, 

226, 230, 231, 232 
Bedford, T., 137, 264 
Benedict, C. G., 41 
Benedict, F. G., 35, 41 
Bentley, M., 97 
Berens, C., 141 
Berger, C., 144 
Berrien, F. K., 129 

Bills, A. G, 42, 68, 69, 78, 87, 99, 100, 

Birren, J. E., 117 
Bitterman, M. E., 47, 87, 102, 141 
Blankenstein, S. S., 115 
Bloch, W., 228 
Bolanovich, D. J., 244 
Borschtschewski, A. S., 151 
Brozek, J., 117 
Brush, A. L., 290 

Cain, P. A, 199, 200, 201, 205 
Carroll, P, Jr., 215 
Chambers, E. G., 273, 278, 285 
Clothier, R. C, 175 

Cobb, P. W., 266, 279, 280, 281, 282, 294 

Cohen, L., 219, 220 

Collier, R. M., 175 

Cooperman, N. R., 154, 155 

Courts, F. A., 104, 109 

Cox, J. W., 306 

Crowden, G. F., 167 

Culpin, M, 190, 293 

Darrow, C. W., 107, 108 
Davis, R. C, 104, 105, 127 
Dealey, W. L., 166 

DeSilva, H. R, 270, 271, 276, 279, 282, 

Dickson, W. J., 150, 178, 181, 192 

Drinker, P., 137 

Dunajewski, M. J., Ill 

Dunbar, H. F., 191, 290, 291, 292, 295 

Dvorak, A., 166 

Enzer, N., 115 
Ewart, E., 244 

Farmer, E, 161, 190, 270, 273, 274, 275, 

278, 285, 287, 295 
Fay, N. J., 112 
Ferderber, M. B., 137 
Fere, C, 53, 58, 60 
Ferree, C. E., 140, 141, 143 
Fisher, M. B., 117 
Fleischer, W. L, 137 
Fletcher, E. D., 272, 273, 283, 284, 294 
Flinn, R. H, 111, 116, 256 
Florence, P. S., 71, 72, 260 
Forbes, T. W., 271, 282 
Ford, G. C, 166 
Eraser, J. A., 73 
Freeman, G. L., 106, 107, 108 
French, L., 110 

Giese, W. J., 108 
Giesecker, F. E., 137 
Gillespie, R. D, 101 
Golden, C. S., 223 
Goldmark, J, 71, 260 
Gordy, C. B., 233, 234, 235 
Gorkin, S. D., 151 
Graf, O., 80, 150, 151 
Guilford, J. P., 245 
Gutberlet, C, 137 

Hammond, E. C, 111, 256 
Harmon, F. L., 127 
Hartmann, G. H., 163 
Heinrich, H. W., 286 
Herdan, G., 287 
Hersey, R. B., 293 
Hoagland, H, 112 
Hoke, R. E., 166 



Holcomb, R. L., 262 
Hollingworth, H. L., 68 
Holway, A. H., 115, 141 
Hopkins, M. D, 71, 72, 260 
Houghten, F. C, 137 
Hunt, J. McV, 191 
Hunt, T., 110 
Hunter, O. B., 110 
Huntington, E., 133 
Hurvich, L. M., 115, 141 

loteyko, J., 51 
Irwin, O. C, 108 

Jellinek, E. M, 262 

Johnson, H. M., 29, 86, 88, 153, 242, 256, 

262, 279, 282, 292 
Jones, B. F, 111, 116, 256 
Juran, J, M, 219 
Jurgensen, C. E., 285 

Kaplan, P. M., Ill 
Keys, A, 117 
King, B. G., 117 
Kirk, F. J., 273, 278 
Kleitman, N., 152, 154, 155 
Knehr, C. A, 141 
Kraepelin, E., 80 
Kurbatova, I. N., 56 
Kustanowitsch, B. J., 151 

Laird, D. A., 127, 154 

Langdon, J. N., 131, 198 

Lauer, A. R., 276, 279 

Lawshe, C. H., 250, 252 

Lee, F. S., 71, 72, 260 

Lee, R. H., 114, 116 

Levina, A. E., 110 

Lewina, A. J., Ill 

Lewitina, G. A., Ill 

Liddell, H. S., 191 

Link, H. C, 73, 74, 199, 254 

Lorge, L, 228 

Louden, J. K, 219 

Lovekin, O. S., 102 

Lowry, S. M., 214, 215, 223, 225 

Luckiesh, M., 138, 139, 140, 141 

McFarland, R. A., 115, 141, 262 
McGeoch, J. A., 304 
Mace, C. A., 171, 172, 173 
Maggiora, A., 58, 60 
Mathewson, S. B., 175 
Maynard, H. B, 214, 215, 223, 225 
Merrick, N. L, 166 

Mills, C. A, 133 
Moore, B. V., 163 
Morgan, J. J. B., 104, 126 
Moss, F. A., 110 
Mosso, A., 49, 56 
Mullin, F. J, 154, 155 
Munsterberg, H., 203 
Murschhauser, H., 35 
Muscio, B., 47 

Newbold, E. M, 266 
Norton, H. F, 165 

Offner, M., 80 
Ogle, C, 133 
Osborne, E. E., 264 

Paterson, D. G., 144 
Pincus, G., 112 

Poffenberger, A. T., 28, 40, 41, 45, 46, 

68, 128, 161, 185 
Pollock, K. W., 233 
Presgrave, R., 214, 216, 221, 223 

Rand, G., 140, 141, 143 
Reid, C., 56, 59 
Richter, W., 150, 151 
Robinson, E. S., 68 
Roe, J. H, 110 

Roethlisberger, F. J., 150, 178, 181, 192 

Rosenberg, A. A., 137 

Rosenblum, D. E,, 41 

Rothe, H. F., 74 

Rounds, G. H., 41 

Rubarth, B., 165 

Ruttenberg, H. J., 223 

Ryan, A. H., 82, 83, 84, 85, 86, 87 

Ryan, T. A., 35, 79 

Samytschkina, K. S., Ill 
Sappenfeld, P. R., 22 
Satter, G. A., 250, 252 
Schubert, H. J. P., 41 
Scott, W. D., 175 
Seashore, S. E., 244 
Segur, A. B., 233 
Shakhnovich, C. E., 110 
Shapiro, S. S., Ill 
Sharp, L. H., 58, 59 
Sheidin, Y. A., 56 
Shumard, F. W., 214, 215 
Sidorowa, L. M., Ill 
Simonson, E., 54, 56, 57, 61, L15 
Simpson, R. M., 107 
Skliankaia, R. M., 110 



Smith, M, 81, 82, 83, 190 
Snell, P. A., 114 
Snow, A. J., 276 
Soloway, E., 102 
Spriegel, W. R., 175 
Stacey, A. E., Jr., 137 
Stegemerton, G. J., 214, 215, 223, 

Stock, F. G. L., 73, 131, 198 
Strauss, L., 219 
Swan, T. H., 153 

Tasker, C., 137 

Tauber, E. S., 290 

Taylor, D. W., 309 

Taylor, F. W., 164 

Thorndike, E. L., 67 

Tiffin, J., 240, 244, 246, 265, 286 

Tinker, M. A., 138, 139, 141, 142, 144 

Titelbaum, S., 154, 155 

Tschernomordik, O. S., Ill 

Vernon, H. M., 73, 127, 145, 257, 258, 

259, 261, 264, 278 
Volmer, E, 117 

Warner, C. G., 127, 264 
Warner, M., 82, 83, 84, 85 
Watson, G., 176, 177 
Wechsler, D., 277, 278 
Weigand, G. E., 153 
Weiss, A. P., 276 
Wenger, M. A., 108 
Wenzig, K., 164 
Weston, H. C., 129 
Wheeler, W. L., Jr., 154 
Wilson, C. H., 112 
Wolfe, T. P., 290 
Woodworth, R. S., 107, 304 
Wyatt, S., 73, 130, 131, 185, 186, 198, 
202, 204, 206, 207, 208, 214 

Yaglou, C. P., 137 

Uhrbrock, R. S., 218 

Zak, A. I., 110 


Absenteeism, as an index of efficiency, 

Absolute validity of time standards, 222- 

Accident control, psychological prob- 
lems in, 253-296 

Accident distribution, 268, 269 

Accident proneness, 264-296 ; clinical use 
of tests for, 281-284; defined, 265; 
generality of trait, 284-289; person- 
ality factors in, 289-293; relation to 
neurosis, 293 ; tests for, 270 ff. 

Accident rates, as measures of efficiency, 
120; comparison with war casualties, 
253 ; prediction by tests, 278-281 

Accident types, relations of, 286-289 

Accidents, alcohol as a factor in, 260-263 ; 
due to sleeping driver, 255, 256 ; emo- 
tional factors in, 289-293 ; fatigue as a 
factor in, 254-260 ; lighting as a factor 
in, 263 ; morning vs. afternoon hours 
in industry, 259 ; personality factors in, 
289-293 ; reduction by educational cam- 
paigns, 254; relation of driving and 
other accidents, 288, 289 ; relation of 
major and minor accidents, 285, 286; 
temperature as a factor in, 263, 

Achievement tests, 299 

Action potentials. See Muscle potentials 

Acuity, visual, effect of level of illumi- 
nation on, 141 

Alcohol, comparison of alcohol and fa- 
tigue effects, 88; effect on habitual per- 
formances, 262 ; relation to accidents, 

Alternation of shifts, their effect on 

production, 149 
Applied psychology, nature of, 3-7 
Aptitude tests, combining scores in, 242 ; 

effort index in, 22. See also Driving 


Arrangement of working place, 159 
Attitudes, as a factor in efficiency, 173, 

174, 186, 187. See also Incentives 
Automobile driving, effects of, 82-87, 111, 

116. See also Accidents 

Barrow work, energy cost of, 167 
Blinking, as an index of effort in visual 
work, 141 

Blocking, as a factor in output decre- 
ment, 69 

Blood count, as related to effort and fa- 
tigue, 110, 111 ; effect of muscular work 
on, 110; effect of sedentary work on, 
110, 111; effect of truck operation on, 

Blood pressure, as an index of effort, 101, 
102; in sedentary work, 101, 102 

Bonus payments, their effect on produc- 
tion, 185, 186 

Boredom, 184, 196-208; effect of music 
on, 130, 131; energy cost of, 42; fac- 
tors causing susceptibility to, 203-208 ; 
generality of trait of susceptibility, 206, 
207; nature of, 197; questionnaire 
studies of, 202; relation of fatigue to, 
197; relation to intelligence, 204; rela- 
tion to physiological measures, 202; 
susceptibility to, 203-208 

Calculation, output decrement in, 68 
Capacity, changes in, related to efficiency, 
89 ; definitions of, 30 ; reduced in fa- 
tigue, 43, 45-48; relation to effort, 31 
Carelessness, tests for, 278 
Central nervous system, in fatigue, 55 
Chemical theories of fatigue, 60 
Circulatory indices of effort and fatigue, 
effect of illumination on, 141 ; in seden- 
tary work, 100-103 
Clinical treatment of accident-prone 

drivers, 281-284 
Cold, effects of. See Temperature 
Color of illumination, effects of, 143 
Comfort chart for effective temperature, 

Comfort reports, related to temperature, 

Comfort zones for temperature, 137 
Conflict, as a factor in work decrement, 

Contrast, visual, in working materials, 
143, 144 



Control, experimental, 16-19 

Convergence reserve, as a measure of 
visual fatigue, 140, 141 

Cooling. See Temperature 

Coordination, effect of fatigue on, 60; 
tests in accident-proneness, 270 ff. 

Correlation, 12-16; curvilinear, 16; 
linear, 13, 14; multiple, 16; signifi- 
cance of, 15, 16; of types of accidents, 

Cost of work, 7, 8 ; definition of, 21 ; 

dissatisfaction as a factor in, 190 ; 

overhead, 40. See also Effort, Energy 

cost of work. Fatigue 
Crank, effect of height and load on 

energy cost of work, 38 
Critical levels of illumination, 139 
Critical scores, 243 

Day shift, compared with night shift, 148 
Decrement. See Output decrement 
Dependence of variables, 12, 16-19 
Depth of fatigue, 53 
Design, of tools and equipment, 159, 164- 

167 ; of typewriters, 165-167 
Difference, significance of, 11 
Discrimination, effect of illumination 

level on, 140 
Disease, effect of temperature on, 133; 

emotional factors in, 191 
Dissatisfaction with work. See Boredom, 

Job satisfaction 
Distraction, effects of, 124. See also 


Distribution, of Illumination, 142; 143; 

of practice, 308 ; of rest periods, 150 
Driver tests, 270-284; validity of, 270- 


Driving, permissible hours, 256, 257. See 

also Accidents, Automobile driving 
Dynamic work. See Work, muscular 

Ease of seeing, design of numbers for, 

144. See also Illumination 
Economy of work. See Efficiency 
Educational campaigns in accident reduc- 
tion, 254 

Effective temperature, chart, 136 ; defini- 
tion, 135 

Efficiency, aims of study, 26-30 ; defini- 
tion of, 7, 8, 20 ; effect of boredom on, 
196-208; effect of incentives on, 169- 
187; effect of lighting on, 138-144; 
effect of loss of sleep on, 154; effect 
of methods of work on, 157-168; effect 

of noise on, 124-131 ; effect of rest 
periods on, 149-151 ; effect of tempera- 
ture on, 131-138 ; ideal of, 28, 29 ; inter- 
relations of factors in, 30, 31 ; misuses 
of term, 21 ; problems of, 7-9 ; relation 
to emergency, 26-28 ; relation to fa- 
tigue, 47 ; relation to muscle tension, 
109; relation to pulse rate, 102; sum- 
mary of the concept of, 32 
Efficiency measures, absenteeism, 121 ; 
accident rates, 120 ; change in capacity, 
89-91 ; health records, 121 ; statistical, 
118-121; turnover rates, 121. See also 
Effort, Energy cost of work. Fatigue, 

Efficiency, mechanical, computation of, 
37; definition of, 20; effect of rest 
periods on, 151 ; measurement of, 
35-42; of lifting by a crank, 38; of 
muscular work, 38-42 ; of pulling and 
pushing, 39 ; of walking and running, 
40; relation to fatigue, 54 

Efficiency, total, 29, 30 

Effort, concept of, 95-98; definition of, 
22; effect of practice on, 297; rating 
of, 215, 216; relation to incentives, 
169 ; relation to time standards, 210 

Effort measures, blood count, 110, 111; 
circulatory, 100-103 ; in visual work, 
138-142; ketosteroid output, 112, 113; 
muscle tension, 103-109 ; pulse product, 
102 ; pulse rate, 100 ; physiological, 95- 

Effort rating, in time standards, 215, 216 
Electrical resistance of the skin, as an 

index of tension, 106-109 
Emergency conditions, efficiency in, 26- 


Emotional factors, in accidents, 290-293 ; 
in disease, 191 ; related to fatigue, 188- 

Emotional stability, in accidents, 276, 277 
Energy cost of work, 21, 35-42 ; effect of 
height and load in lifting by a crank, 
38; effect of rest, 151; effect of shovel 
design on, 164; of barrow work, 167; 
of sedentary work, 41-42 ; of work dur- 
ing noise, 127-128 
Ergograph, 49 

Ergographic experiments, results of, 

Errors, related to efficiency, 120 
Exhaustion, 52; sudden, following slow 

work, 60 
Exit interview, 174 


Factorial analysis, of accident types, 287 
Factory studies, of effect of noise, 129, 
130; of incentives, 173-176; of lighting, 
139; of rest periods, 149-151. See also 
Efficiency, Output, Output curves 
Factory work, incentives in, 169-187; 
output decrement in, 70-75 ; repetitious, 

Fatigue, and mechanical efficiency, 54 ; 
as a cause of accidents, 254-260 ; cy- 
clical character of recovery from, 58 ; 
defined as reduced efficiency, 47 ; defi- 
nitions of, 42, 43 ; depth of, 53 ; effect 
of, on the blood count, 110, 111 ; emo- 
tional factors in, 188-208; method of 
computing decrement, 67; output de- 
crement in factory work, 70; physio- 
logical basis of, 44 ; relation to alcohol 
effects, 88 ; relation to boredom, 197 ; 
transfer of, 78, 87, 88 

Fatigue allowances, 213, 225-227; table 
of, 226 

Fatigue decrement. See Output decre- 

Fatigue in muscular work, 48-62 ; chem- 
ical basis of, 60, 61 ; effect of load on 
rate of development of, 60 ; effect of 
rate of work on, 60; effect of, on co- 
ordination, 60 ; lactic acid in, 61 ; loci 
of, 55-57; spread of activity in, 60 

Fatigue in sedentary work, 63-95 ; de- 
crement resulting from, 65-75 ; methods 
of study of, 75-78 

Fatigue measures, flicker-fusion fre- 
quency, 113-118; ketosteroid output, 
112-113; physiological tests, 95-123; 
reduced capacity in, 43, 45-48 ; reduced 
efficiency in, 47; work decrement, 48, 

Fatigue, nervous, 188-208 

Fatigue tests, 76-88; effects of automo- 
bile driving on, 82-87 ; effects of sleep- 
loss on, 81-83, 153 ; in finding per- 
missible hours of driving, 256, 257 

Flicker-fusion frequency, 113-118; effect 
of exposure to flicker on, 114, 115; 
effect of heavy muscular work on, 117 ; 
effect of low oxygen pressure on, 
117; effect of sleep-loss on, 115; effect 
of truck driving on, 116 

Focusing, speed of, as affected by illumi- 
nation, 140 

Food, effect of, on sleep, 154-156 

Force, amount required as a criterion of 
typewriter design, 165; dependent on 

handle design, 165 ; mechanical, in mus- 
cular work, 33 
Fuel consumption. See Energy cost of 

Generality of traits, accident-proneness, 

284-289; susceptibility to boredom, 

Glare, effect of, 138, 142-143 
Goal, effect of, on output; 171-173. See 

also Time standards 
Governing factors, 124-156 relation to 

incentives, 124 
Graphic rating method, 238-246 

Halo effect, in industrial merit rating, 

Handles, design of, for screwdriver, 165 
Health, effect of temperature on, 133; 
emotional factors in, 191 ; records as a 
measure of efficiency of, 121 
Heart rate. See Circulatory indices, 

Pulse rate 
Heat, effects of. See Temperature 
Homogeneity, as a factor in work de- 
crement, 68-70; relation of, to motion 
study, 161 

Hours of work, 145-149; effect on acci- 
dents, 257-259; permissible, in truck 
operation, 256, 257 

Humidity, effect of, 132 

Illumination, color of, 143 ; critical levels 
of, 139 ; effect of level of, on acuity, 
143; effect of level of, on rate of 
blinking, 141 ; effect of level of, on 
speed of discrimination and focusing, 
140; glare, 142-143; indirect, 143; 
levels of, 138-142; methods of deter- 
mining optimal levels of, 138-142 ; pref- 
erences for, 141-142; relation of, to 
accidents, 263 

Impairment of efficiency, relation to fa- 
tigue tests, 86 

Impairment of output. See Output de- 

Incentives, 8, 169-187; effect of time 
standards on, 209, 224; in learning, 
299, 309, 310; methods of study of, 
170-176; pay, 8, 9, 175, 183-186, 214; 
relation of, to efficiency, 169; relation 
of, to governors, 124, 169 ; relation of, 
to merit rating and job evaluation, 
237 ; results of questionnaire studies 
on, 176-178; social factors of, 186; 



Incentives (cont'd) 
standards and goals for, 171-173, 182; 
Western Electric Company studies of, 
150, 178-182 

Incentives, industrial, field studies of, 
173-176; naturalistic studies of, 178 

Individual differences, in accident lia- 
bility, 264-296 ; in studies of efficiency, 
39 ; relation of, to motion study, 162, 
163 ; relation of, to synthetic time 
standards, 233-235 ; relation of, to time 
standards, 213, 214; susceptibility to 
monotony, 196, 203-206. See also Acci- 
dent proneness 

Input, definition of, 21. See Cost of work. 
Effort, Energy cost of v^ork 

Intelligence, relation of, to accidents, 
270, 278 ; relation of, to boredom, 203, 
204; relation of, to turnover rate, 175 

Intensity of light. See Illumination, 
levels of ; Lighting 

Intent to learn, 309 

Interest, 183-186; as a factor in learn- 
ing, 310 

Interference in learning, 305-308 
Intersection accidents, and vision, 276, 

Interview studies of incentives, 174, 179 

Job analysis. See Job evaluation. Meth- 
ods of work, Motion study 

Job evaluation, 209, 237, 238, 250-252; 
factors in, 251 ; halo effect in, 252 

Job hazards, differences in, 265 

Job preferences, 177 

Job satisfaction, 177, 178, 184 

Ketosteroids, as an index of fatigue, 112, 

Lactic acid, role of, in fatigue, 61 
Learning, adjustive, 302 ff.; by wholes, 
304; effect of sequence on, 303-308; 
effect of, on muscle tension, 109 ; effi- 
ciency of, 297-312; incentives and mo- 
tives in, 309-310; interest as a factor 
in, 310; interference, 305-308; kinds of, 
301-303 ; memorizing, 302 ff. ; motor, 
302 ff. ; relation of, to comprehension, 
302 ff. ; sequence of material, 303, 304 ; 
understanding as a factor in, 310, 311 
Legibility of type faces, 144 
Length of rest periods, effect of, 58, 150 
Length of working day, effect of, on 
accidents, 259; effect of, on produc- 
tivity, 146 

Leucocytosis. See Blood count 
Leveling factors, table of, 215 
Leveling method. See Time standards 

with rating 
Levels of effort, criteria of, 223 
Lighting, effects of, 138-144. See also 

Load, effect of, on efficiency, 38, 39; 

effect of, on fatigue, 60; optimal, for 

shoveling, 164; relative, on two hands 

in typing, 166 
Locomotion, mechanical efficiency and 

rate of, 40 

Machine work, output curves in, 70-75 
Machines, design of, 164-167 
Major and minor accidents, relations of, 
285, 286 

Mechanical efficiency. See Efficiency, me- 
chanical. Energy cost of work 

Mechanical recording, of muscle tension, 

Memorizing, factors in, 302 ff. 
Mental work. See Work, sedentary 
Merit rating, 237-250; correcting for 

differences of standards, 245 ; halo 

effect in, 244 ff. ; reliability of, 240 ff. ; 

suggested procedure for, 246-249; 

validity of, 240 ff. 
Metabolic cost of work. See Energ>^ cost 

of work 

Methodology, in applied psychology, 
9-19 ; of setting time standards, 210-217 

Methods engineering. See Methods of 
work, Motion study 

Methods of work, 157-168; arrangement 
of tools and material. 159; design of 
tools, 159, 164-167; principles of. 158, 
159; relation to individual differences, 
162, 163; use of the body in, 158 

Micromotion chart, 158 

Monotony. See Boredom 

Morale, 169 

Motion economy. See Methods of work 
Motion study, 157-168; criticisms of, 
160-164; relation to efficiency, 160. 
See also Methods of work 
Motivation, problems of, 8. See also In- 

Motor learning, factors in, 302 ff. 

Motor skill, 64 ; concept of, in time stand- 
ards, 211, 212. See also Learning 

Muscle potentials, 104-106; in rest. 58. 59 

Muscle tension, as an index of effort, 
103-109; effect of learning on, 109; 



effect of lighting on, 141 ; effect of 
noise on, 104, 126-128; effect of, on 
output, 109; relation of, to efficiency, 
109; relation of, to electrical skin re- 
sistance, 106-109; role of, in sedentary 
work, 99, 100, 103-109 
Muscular work. See Work, muscular 
Music, effects of, 130, 131; set as a 
factor in determining effect of, 131 

Nervous fatigue, 188-208 

Neuroses, experimental, 191 ; occupa- 
tional, 190; relation of, to accident- 
proneness, 293 

Night shifts, effect of, on productivity, 

Noise, effect of, on efficiency and pro- 
ductivity, 124-131 ; effect of, on muscle 
tensions, 104, 126-128 ; effect of, on the 
energy cost of work, 127, 128 ; field 
studies of its effects in industry, 129, 

Ocular work, factors in, 138-144 
Output, effects of boredom on, 197-202; 
effect of emotional factors on, 188, 189 ; 
effect of illumination on, 139; effect 
of loss of sleep on, 153 ; effect of 
muscle tension on, 109 ; effect of noise 
on, 125, 129 ; effect of rest periods on, 
149; effect of temperature on, 133; 
measurement of, in foot-pounds, 36; 
relation of, to capacity, 47 ; relation of, 
to tiredness, 46; restriction of, 175, 
176; under the piece-rate system, 185, 
186, 214; units of measurement of, 
37, 38 

Output curves, in boring work, 197-201 ; 
in factory work, 70-75; in inspection 
work, 73, 74, 199 

Output decrement, as a fatigue measure, 
45-48; effect of conffict in task on, 
68-70; effect of homogeneity of task 
on, 68-70; factors in, 65-70; in factory 
work, 70-75; in a packing operation, 
74 ; in sedentary work, 48, 65-75 ; 
method of computing, 67 ; related to 
fatigue allowances, 225 

Output rate, relation to accidents, 258- 

Output trends, relation to efficiency, 119 
Overlapping motions, in synthetic time 

standards, 233 
Oxygen consumption, as a measure of 

the energy cost of work, 35 

Oxygen debt, 36 

Oxygen deprivation, effects of, on fficker- 
fusion frequency, 117 

Part vs. whole learning, 304 

Pay incentives, relation of, to motion 
study, 161. See also Incentives 

Perceived fatigue. See Tiredness 

Performance rate. See Output 

Personal factors. See Accident-prone- 
ness ; Boredom, susceptibility to ; Indi- 
vidual differences 

Personality, and accidents, 289-293; and 
boredom, 203-208 

Physiological effects, of boredom, 202 ; 
of sleep, 152 ; of work, 95-123 

Piece-rate methods, effect of, on pro- 
duction, 185, 186, 214 

Practice, effect of, on effort, 297; effec- 
tive, 297-312; scheduling of, 299, 308, 
309 ; sequence of, 303-308 ; size of 
unit of, 304, 305. See also Learning 

Practice effect, control of, in measuring 
fatigue decrement, 67 

Prediction, 12-16 

Problem-solving, 64 

Process chart, 158 

Production rate. See Output 

Propaganda, effect of, on accident rates, 
254 ; methods of evaluating, 254 

Psychiatric study of accident-proneness, 

Psychology, applied, the nature of, 3-7 
Pulling vs. pushing, relative energy cost 
of, 39 

Pulse-product, as an index of effort, 102 
Pulse rate, as an index of effort, 100- 
102; in visual tasks, 102 

Questionnaire, studies of boredom, 202; 
survey of incentives, 173, 174, 176-178 

r (symbol for linear correlation). See 

Radiation, as a factor in temperature 

control, 132 
Rate of work, problems of, 26-28 ; effect 

of, on fatigue and output, 53, 60 
Rate setting. See Job evaluation. Time 


Rating methods, in time study, 213-236. 

See also Merit rating, Job evaluation 
Reaction time, in accident-proneness, 

270 ff. 
Recollection, 301 



Recovery, cyclical character of, 58 ; mus- 
cle tension during, 58, 59. See also 

Reliability, of merit ratings, 240 ff. ; of 
time standards, 217-221 

Repetitive work, effects of. See Bore- 

Research methods, general, 9-19 
Resistance of skin. See Electrical re- 
sistance of skin 
Rest, following muscular work, 57-59 ; 

muscle potentials during, 58, 59 
Rest periods, 145, 149-151; effect on 
pulse-product, 103 ; in sedentary work, 
150, 151; optimal length, 150; West- 
ern Electric Company studies of, 150 
Restriction of output, 175, 176 
Retroactive inhibition, 307, 308 
Running, mechanical efficiency of, 40 

Safety propaganda, problem of effec- 
tiveness of, 254 

Sampling error, in correlation, 15, 16 

Satisfaction with work. See Job satis- 

Scatter diagram, 14 

Scheduling of practice, 299, 308, 309 

Screwdrivers, design of, 165 

Sedentary-muscular work, 64 

Sedentary work. See Work, sedentary 

Selection, vocational, 8, 9 

Sensory adjustment, 64 

Sequence, effect of, on learning, 299, 

Shifts, alternating, 149; day and night, 

Shovels, design of, 164; effect of design 
of, on energy cost of work, 164, 165 ; 
optimal load of, 164, 165 

Significance, of correlation, 15, 16; prac- 
tical, 10-12; statistical, 10-12, 15, 16 

Skill, development of, 297-312; relation 
of, to time standards, 211. See also 

Skin resistance. See Electrical resistance 
of skin 

Sleep, 145, 151-156; criteria of, 152; dis- 
tribution of, 152; effect of food on, 
154-156; effect of loss of, on energy 
cost and efficiency, 154; effect of sleep 
loss on test performance, 81-83, 153 ; 
movement during, 154; physiological 
changes during, 152 

Sleeping drivers, as a cause of accidents, 
255, 256 

Social factors in work, 186 
Spacing, of muscular contractions, effect 
of, 60 

Speed of work. See Output rate 
Spread of muscular activity in fatigue, 

Standards, effect of, on output, 171-173, 
183; methods of determining. See 
Time standards 
Static work. See Work, muscular 
Statistical measures, of efficiency, 118- 

Statistical methods, 10-16; of setting 
time standards, 228 

Supervisory relations, 186 

Susceptibility to boredom. See Boredom 

Synthetic time standards, 229-235 ; rela- 
tion to individual differences, 233-235 

Telegrapher's cramp, 190 

Telegraphic code, learning of, 309 

Temperature, comfort zones, 137; effect 
of, 131-138; effect of, on health, 133; 
methods of determining optimum, 132- 
138; physiological effects of, 132; rela- 
tion to accidents, 263, 264 

Tests, achievement, 299. See also Acci- 
dent proneness, Fatigue tests 

Therblig, definition of, 158 

Time standards, 209-236; leveling 
method for, 213-236; methods of set- 
ting, 210-217; rating methods for, 213- 
236 ; relation of, to average time. 214, 
228-229; relation of, to individual dif- 
ferences, 213, 214; reliability of, 217- 
221 ; statistical methods of setting, 228 ; 
synthetic methods for, 229-235; tra- 
ditional, 212, 213 ; use of several rates 
of, for improving reliabilitv, 219; 
validity of, 221-227 

Time study, as a method of determining 
the design of equipment, 164. See also 
Time standards 

Tiredness, feelings of, 43-45 ; in relation 
to output, 46 

Tonus. See Muscle potentials, Muscle 

Tools, design of, 159, 164-167 

Training, effective, 297-312; of merit 
raters, 244 

Transfer, of fatigue, See Fatigue; of 
training, 305-308 

Truck driving. See Accidents, Automo- 
bile driving 

Turnover rate, analysis of, 174; as in- 


dicating efficiency, 121 ; relation of, to 
intelligence, 175 
Types of accident, interrelations of, 286- 

Typewriters, design of, 165, 166 

Understanding, as a factor in learning, 
310, 311 

Uniformity of task, its effect on decre- 
ment, 68-70. See also Boredom 

Validity, of driver tests, 270-281; of 
merit ratings, 240 ff. ; of time stand- 
ards, 221-227 

Ventilation, effects of, 131-138 

Visibility, of numbers, 144 

Visual tests, in accident-proneness, 
270 ff. ; relation to intersection acci- 
dents, 276, 294 

Visual work, factors in, 138-144 

Vocational selection, 8, 9 

Wages. See Incentives 
Walking, mechanical efficiency of, 40 
Warming-up, effect on work-curves, 68 
Weighting, of traits in rating, 243 ff. 
Western Electric Company researches, 
150, 178-182, 192 

Wheelbarrow, design of, 167; work with, 
energy cost of, 167 

White cells. See Blood count 

Whole learning, 304 

Work, classification of, 33-35; dromal 
tasks, 57; dynamic, fatigue in, 55, 56; 
static, fatigue in, 56, 57 

Work, muscular, definition of, 33 ; effect 
on blood count, 110; effect on fatigue, 
48-62 ; effect on flicker-fusion fre- 
quency, 117; mechanical efficiency of, 
38-42 ; static vs. dynamic, effects of, 
55-57 ; types of, 35 

Work, sedentary, circulatory indices of 
effort in, 100-103; definition of, 33; 
effect of noise on, 124-131 ; effect of 
rest periods on, 150 ; effect of tempera- 
ture on, 133 ; effect on blood count, 
110, 111; effect on flicker-fusion fre- 
quency, 113-118; effect on pulse-rate 
and pulse-product, 100-102; energy 
cost of, 41, 42; fatigue in, 63-95 ; kinds 
of, 63-65 ; muscle tension during, 103- 
109 ; output decrement in, 48 ; physio- 
logical correlates of, 95-123; rest 
periods in, 150, 151 ; role of muscular 
activity in, 99, 100 

Working week, optimal length of, 146 

Date Due 

%?R2§ 55 


if- ■ 

MAS 1 


MAR i 


L-- - 

NOV 2 § M 


DEC 1 7 -f 



3 1262 04810 8327